February 2005 - Darwin-kernel

Re: another strange panic
by Andrew Gallatin 28 Feb '05

28 Feb '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Mark Cookson writes: > On 2/28/05 4:13 PM, "Andrew Gallatin" <gallatin(a)cs.duke.edu> wrote: > > > Sound assertion "!success && ( 0x20 == address )" Failed in > > AppleOnboardAudio/K2Platform.cpp line 285 goto Exit > > That particular error is the built in audio driver not being able to talk > over I2C to the sound chip (0x20 is the base address of the chip's > registers). Do you do anything with the I2C driver that might wedge it? No. Our driver is for a PCI device. > This call path is in the initialization of the driver (are you unloading > it?), but it should only be retried one or two times, not infinitely. > > Unfortunately, I can't think of anything that this might mean in relation to > a third party driver, unless you're actually loading and unloading the > AppleOnboardAudio driver. We never unload the sound driver. Just our own driver. Hmm.. I wonder if we've got a bogus stack someplace caused by some sort of stack overflow, and the garbage return address just happens to match the address of this audio driver init routine.. Drew _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: another strange panic
by Mark Cookson 28 Feb '05

28 Feb '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com User-agent: Microsoft-Entourage/11.1.0.040913 On 2/28/05 4:13 PM, "Andrew Gallatin" <gallatin(a)cs.duke.edu> wrote: > Sound assertion "!success && ( 0x20 == address )" Failed in > AppleOnboardAudio/K2Platform.cpp line 285 goto Exit That particular error is the built in audio driver not being able to talk over I2C to the sound chip (0x20 is the base address of the chip's registers). Do you do anything with the I2C driver that might wedge it? This call path is in the initialization of the driver (are you unloading it?), but it should only be retried one or two times, not infinitely. Unfortunately, I can't think of anything that this might mean in relation to a third party driver, unless you're actually loading and unloading the AppleOnboardAudio driver. Very strange indeed. -- Mark Cookson M-Audio, a division of Avid 225 Locust St. Hudson, WI 54016 _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

another strange panic
by Andrew Gallatin 28 Feb '05

28 Feb '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Every few weeks, one of our tinderbox machines "hangs." These tinderbox machines load and unload our driver many times per day. I've captured one clean crash, but mostly the machine hangs with messages scrolling on the screen. Nothing is logged in the panic log for the hang, nor can I attach a remote gdb to it. I finally got the guy on site to get a camera and take a picture when it happened this morning. The following was scrolling on console: <.... repeat ... > Sound assertion "!success && ( 0x20 == address )" Failed in AppleOnboardAudio/K2Platform.cpp line 285 goto Exit Exception state (sv=0x00337230) PC=0x000924A0; MSR=0x0001030; DAR=0x003373D8; DSISR=0x003373E0;LR=0x003373C8; R1=0x003372B8; XCP=0x003373E8 (Unknown code) Backtrace: backtrace terminated - unaligned frame address: 0x003372B8 <.... repeat ... > Note that this was copied by hand from a blurry jpeg, so I may have mis-transcribed "8" vs "B" vs "0". Please also note that I have no way to know if the audio assert came before or after the attempted backtrace. The only information I can extract is that the PC seems to match 000924a0 T _switch_in. The "other" state that things end up in is with a crash in do_thread_scan() due to a corrupt processor state (see "Re: strange panic, debugging help wanted.." from feb 16th). Could this crash also be due to a corrupt processor state? But one whose garbage data was good enough to actually get to the point of trying to run a thread? I'm still working on the assumption that this must be my driver's fault. Any idea what I might have scribbled over to trigger such a condition? I do some unusual things, like starting and stopping threads via IOCreateThread()/IOExitThread(), and killing processes via pfind() and psignal(). I'm sorry to be so vague, but I don't have very much to go on and I was hoping somebody might have fought something similar. Thanks, Drew _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: KEV_DL_IF_DETACHED/KEV_DL_IF_ATTACHED: how to?
by Stephane Sudre 28 Feb '05

28 Feb '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On 28 févr. 05, at 18:35, Josh Graessley wrote: On Feb 28, 2005, at 5:20 AM, Stephane Sudre wrote: On 28 févr. 05, at 13:01, Stephane Sudre wrote: On 26 févr. 05, at 02:52, Justin Walker wrote: On Feb 25, 2005, at 8:36, Stephane Sudre wrote: Thanks. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… What kind of operation/action can trigger a KEV_DL_IF_ATTACHED or KEV_DL_IF_DETACHED kernel event? If you haven't tried it yet, consider getting a copy of the xnu kernel source and using 'cscope'. This will let you track down the answer to such questions (as long as the answers lie in the BSD/Mach part of the kernel; cscope does not grok C++ [and it's not alone]). Will have a look but I'm not sure it's going to tell me how I can physically reproduce it. According to the dlil.c code I'm looking at, it looks like that when all proto have been detached (KEV_DL_PROTO_DETACHED) from the interface, the interface is detached (KEV_DL_IF_DETACHED). On the OS version for which I'm trying to see a KEV_DL_IF_DETACHED event , when I disable an interface from the Network Preferences Pane, I see 2 KEV_DL_PROTO_DETACHED events but no KEV_DL_IF_DETACHED event as far as I can tell. KEV_DL_IF_DETACHED is generated when an interface goes away. This happens when the driver call dlil_if_detach. Most driver's don't detach the interface (built-in, airport, PCI). PC Card drivers do detach the interface. PPP creates interfaces on the fly. You will probably see a number of these events for PPP interfaces. You can also trigger these events by creating and removing vlan interfaces using ifconfig. This email sent to site_archiver(a)lists.apple.com

1 0

Re: KEV_DL_IF_DETACHED/KEV_DL_IF_ATTACHED: how to?
by Josh Graessley 28 Feb '05

28 Feb '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Feb 28, 2005, at 5:20 AM, Stephane Sudre wrote: On 28 févr. 05, at 13:01, Stephane Sudre wrote: On 26 févr. 05, at 02:52, Justin Walker wrote: On Feb 25, 2005, at 8:36, Stephane Sudre wrote: _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com What kind of operation/action can trigger a KEV_DL_IF_ATTACHED or KEV_DL_IF_DETACHED kernel event? If you haven't tried it yet, consider getting a copy of the xnu kernel source and using 'cscope'. This will let you track down the answer to such questions (as long as the answers lie in the BSD/Mach part of the kernel; cscope does not grok C++ [and it's not alone]). Will have a look but I'm not sure it's going to tell me how I can physically reproduce it. According to the dlil.c code I'm looking at, it looks like that when all proto have been detached (KEV_DL_PROTO_DETACHED) from the interface, the interface is detached (KEV_DL_IF_DETACHED). On the OS version for which I'm trying to see a KEV_DL_IF_DETACHED event , when I disable an interface from the Network Preferences Pane, I see 2 KEV_DL_PROTO_DETACHED events but no KEV_DL_IF_DETACHED event as far as I can tell. KEV_DL_IF_DETACHED is generated when an interface goes away. This happens when the driver call dlil_if_detach. Most driver's don't detach the interface (built-in, airport, PCI). PC Card drivers do detach the interface. PPP creates interfaces on the fly. You will probably see a number of these events for PPP interfaces. You can also trigger these events by creating and removing vlan interfaces using ifconfig. -josh smime.p7s

1 0

Re: Matching and loading filter scheme driver other than "Content Hint" property
by Dan Markarian 28 Feb '05

28 Feb '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi Yogesh, Dan On 25 Feb 2005, at 4:37 AM, Yogesh P wrote: Thanks in advance, Yogesh Pahilwan _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… You are welcome to use other properties that are applicable. You want to limit the number of media objects the filter scheme matches, as a general rule, so be as specific as possible. I am not sure what your concerns might be without more context. Hi Folks, I had write the sample filter scheme driver long back, and it works fine by matching on the "Content Hint" property.But because of some concerns i do not want to load my filter scheme driver by matching on "Content Hint" property. Is there any way or some property other than "Content Hint" on which my filter scheme driver gets match and loads??? Or it is the only property for loading any kind of filter scheme driver???? _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/ markarian%40apple.com This email sent to markarian(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: KEV_DL_IF_DETACHED/KEV_DL_IF_ATTACHED: how to?
by Stephane Sudre 28 Feb '05

28 Feb '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On 28 févr. 05, at 13:01, Stephane Sudre wrote: On 26 févr. 05, at 02:52, Justin Walker wrote: On Feb 25, 2005, at 8:36, Stephane Sudre wrote: _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… What kind of operation/action can trigger a KEV_DL_IF_ATTACHED or KEV_DL_IF_DETACHED kernel event? If you haven't tried it yet, consider getting a copy of the xnu kernel source and using 'cscope'. This will let you track down the answer to such questions (as long as the answers lie in the BSD/Mach part of the kernel; cscope does not grok C++ [and it's not alone]). Will have a look but I'm not sure it's going to tell me how I can physically reproduce it. According to the dlil.c code I'm looking at, it looks like that when all proto have been detached (KEV_DL_PROTO_DETACHED) from the interface, the interface is detached (KEV_DL_IF_DETACHED). On the OS version for which I'm trying to see a KEV_DL_IF_DETACHED event , when I disable an interface from the Network Preferences Pane, I see 2 KEV_DL_PROTO_DETACHED events but no KEV_DL_IF_DETACHED event as far as I can tell. This email sent to site_archiver(a)lists.apple.com

1 0

Re: KEV_DL_IF_DETACHED/KEV_DL_IF_ATTACHED: how to?
by Stephane Sudre 28 Feb '05

28 Feb '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On 26 févr. 05, at 02:52, Justin Walker wrote: On Feb 25, 2005, at 8:36, Stephane Sudre wrote: Detaching the cable, just says the Link is down. Noted. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… What kind of operation/action can trigger a KEV_DL_IF_ATTACHED or KEV_DL_IF_DETACHED kernel event? If you haven't tried it yet, consider getting a copy of the xnu kernel source and using 'cscope'. This will let you track down the answer to such questions (as long as the answers lie in the BSD/Mach part of the kernel; cscope does not grok C++ [and it's not alone]). Will have a look but I'm not sure it's going to tell me how I can physically reproduce it. Ejecting/inserting a WiFi PCMCIA card in an old PowerBook for instance? It probably depends on exactly what the device handler and the corresponding IONetwork family support does. I believe that, for the Ethernet family, insertion/ejection of the card (or a cable) will cause such events, but the code tells the real story. Disabling the Interface in the Network Preference Pane leads to 1 or multiple Detach Protocol message(s). Note that the IONetworkFamily code is a separate module, and is packaged separately in the darwin repository. This email sent to site_archiver(a)lists.apple.com

1 0

Re: <template> classes for kernel development
by Dean Reece 28 Feb '05

28 Feb '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hope this helps, - Dean They make a lot of use of <template> classes. Thanks _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… Templates in the kernel are not explicitly prevented from working, but they are not officially supported. This means that we do not test the compiler or the kernel C++ runtime for templates, nor do we provide technical support to developers for their use. We also don't guarantee binary compatibility between releases w.r.t. templates. That being said, I have heard of developers successfully using them. In general, you can use language feature that you can get to work, as long as it is entirely internal to your kext. Anything that requires special linker or run-time support that we don't officially support is at risk. Officially, we support the same language features as The Embedded C++ (http://www.caravan.net/ec2plus/) The Mac OS X kernel C++ environment is described at http://developer.apple.com/documentation/DeviceDrivers/Conceptual/ WritingDeviceDriver/CPluPlusRuntime/chapter_2_section_1.html On Feb 27, 2005, at 2:52 PM, Alun Carp wrote: I have a bunch of classes which provide support for various encryption algorithms. My question is whether template classes can be used for Mac OS X kernel driver development? Alun This email sent to site_archiver(a)lists.apple.com

1 0

<template> classes for kernel development
by Alun Carp 27 Feb '05

27 Feb '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi, They make a lot of use of <template> classes. Thanks Alun _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… I have a bunch of classes which provide support for various encryption algorithms. My question is whether template classes can be used for Mac OS X kernel driver development? This email sent to site_archiver(a)lists.apple.com

1 0