September 2003 - Darwin-kernel

Custom logging user client tool from GMAC?
by Michael Cashwell 09 Sep '03

09 Sep '03

Greetings, Sorry if this should go to Darwin-drivers. (Some questions span these lists' domains.) I'm trying to make use of the custom logging code (ELG) from Apple's GMAC driver in my own code. It's discussed at http://developer.apple.com/documentation/DeviceDrivers/Conceptual/ WritingDeviceDriver/DebuggingDrivers/chapter_8_section_5.html which says: > This driver implements a kernel-resident logging infrastructure as > well as a user-space tool to print the formatted contents of a log > buffer. The tool used by AppleGMACEthernet prints a report to the > standard output that begins with header information followed by > time-stamped log entries. However, the AppleGMACEthernet-124.2.3.tar archive does not appear to contain any source for the mentioned userland tool. Does anyone know where I might find this source? Thanks, -Mike _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

DMA questions (invalidate_dcache and flush_dcache)
by Ryan Rempel 08 Sep '03

08 Sep '03

I'm updating a driver for the new memory mapping limitations, and I've got a couple of DMA-related questions. The driver uses a couple of functions in <osfmk/ppc/pmap.h>, namely: extern void invalidate_dcache(vm_offset_t va, unsigned length, boolean_t phys); extern void flush_dcache(vm_offset_t va, unsigned length, boolean_t phys); The calls to invalidate_dcache are accompanied by this comment "Purge cache references for the DBDMA entry we are about to look at". The calls to flush_dcache are accompanied by this comnment "Push DMA commands to physical memory". In both cases, the "phys" argument is set to 0. What I'm wondering is whether the recent memory management changes would affect this kind of code. Or, indeed, whether there is a different or better way to accomplish what this code is accomplishing. _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Call prepare before mapDeviceMemoryWithIndex?
by Ryan Rempel 08 Sep '03

08 Sep '03

Is it now necessary to prepare () an IOMemoryDescriptor before calling mapDeviceMemoryWithIndex? That is, where one had, in the past, been doing something like this: IOMemoryMap *mem = provider->mapDeviceMemoryWithIndex (0); Is it now necessary to do something like this: provider->getDeviceMemoryWithIndex (0)->prepare (); IOMemoryMap *mem = provider->mapDeviceMemoryWithIndex (0); Also, is it important to know whether somebody else has already called prepare () on the IOMemoryDescriptor? That is, is there any problem in calling it twice (if, for instance, a superclass has already called it, or that sort of thing). Also, where the memory mapping is going to be used for the entire life of the kernel extension, and the kernel extension cannot be unloaded, is it important to call IOMemoryDescriptor::release at some point? What would be the appropriate point--I suppose during stop ()? _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: Linking question and KPI
by Dean Reece 08 Sep '03

08 Sep '03

Stiphane, The implementation of KPIs is still in its early stages. At this point we are still evaluating our options and are still working out our plan for completing and delivering them. [Also, it is generally not our policy to comment on the contents of unreleased/future software on a public forum.] As the plans and implementation firm up, we will be communicating the details to our developers. We realize that such transitions take time for our developers to complete, so we'll try to provide the information as early as we reasonably can. Cheers, - Dean On Monday, September 8, 2003, at 02:05 AM, Stiphane Sudre wrote: Would this kind of evolution be included in a major or a minor update? On vendredi, sep 5, 2003, at 22:17 Europe/Paris, Joshua Graessley wrote: On Sep 5, 2003, at 12:11 PM, Sean McBride wrote: Does anyone know what's up? Is the whole KPI thing is not baked yet? I suspect these kpi things are not fully working yet. The seed is just of the headers. The symbols are not in the builds of Panther we've been seeding to developers. In addition, the symbols will not be available in Panther when it is released. [...] _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: Linking question and KPI
by Stéphane Sudre 08 Sep '03

08 Sep '03

Would this kind of evolution be included in a major or a minor update? On vendredi, sep 5, 2003, at 22:17 Europe/Paris, Joshua Graessley wrote: On Sep 5, 2003, at 12:11 PM, Sean McBride wrote: Does anyone know what's up? Is the whole KPI thing is not baked yet? I suspect these kpi things are not fully working yet. The seed is just of the headers. The symbols are not in the builds of Panther we've been seeding to developers. In addition, the symbols will not be available in Panther when it is released. [...] _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: Linking question and KPI
by Joshua Graessley 05 Sep '03

05 Sep '03

On Sep 5, 2003, at 12:11 PM, Sean McBride wrote: >> Does anyone know what's up? Is the whole KPI thing is not baked yet? > > I suspect these kpi things are not fully working yet. The seed is just of the headers. The symbols are not in the builds of Panther we've been seeding to developers. In addition, the symbols will not be available in Panther when it is released. I believe the plan is to continue to evolve the KPIs as work on finer grained locking progresses. I think we're hoping to seed some builds as we make more progress. Right now, everyone is focused on shipping Panther, so the KPIs are on the back-burner for now. We do have an implementation of the symbols in the KPIs. It's half baked and requires some changes to the kernel that were deemed to risky. In addition, the KPIs don't account for the locking model, so they're not stable. Instead of releasing KPIs that would change when we changed the locking model, we opted to wait but seed the headers so we could get feedback on the direction we're going with the KPIs. -josh [demime 0.98b removed an attachment of type application/pkcs7-signature which had a name of smime.p7s] _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: Linking question and KPI
by Sean McBride 05 Sep '03

05 Sep '03

Michael Cashwell (mboards(a)prograde.net) wrote on Fri, 5 Sep 2003 12:38:02 -0400: >I'm having link errors in a device driver I'm working on at kextload >time. Due to NDA I can't say what Apple OS or Dev tools I'm running ( >:-p ) I'm having similar problems. :( >I decided to try to use the KPI >stubs. The talk on them at WWDC was interesting and I figured I should >not start a new project using deprecated interfaces. I haven't heard of these until now, but have played with them since you mentioned it. What are they? I assume it relates to "Kernel Dependency Changes" discussed at: <http://developer.apple.com/technotes/tn/tn2090.html> >Any thoughts would be appreciated. Alas, I tried adding every com.apple.kpi.* to my OSBundleLibraries and still I get link errors: 17 'undefined symbols', though all are functions I call explicitly. Does this mean I should not be using these functions at all? Most are benign, like strcpy()! >Does anyone know what's up? Is the whole KPI thing is not baked yet? I suspect these kpi things are not fully working yet. -- ____________________________________________________________ Sean McBride, B. Eng Sean.McBride(a)matrox.com Mac Software Designer +1-514-822-6000 x7399 Matrox Electronic Systems Ltd. Montrial, Quibec, Canada _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Linking question and KPI
by Michael Cashwell 05 Sep '03

05 Sep '03

Greetings all, I'm having link errors in a device driver I'm working on at kextload time. Due to NDA I can't say what Apple OS or Dev tools I'm running ( :-p ) but since my intended deployment OS will be post 10.2 (though I likely could back that down to 10.2) I decided to try to use the KPI stubs. The talk on them at WWDC was interesting and I figured I should not start a new project using deprecated interfaces. So I initially had these OSBundleLibraries: com.apple.iokit.IONetworkingFamily 1.3.1 com.apple.iokit.IOPCIFamily 1.4.0 com.apple.kpi.iokit 7.0.0 com.apple.kpi.libkern 7.0.0 But kld failed with an unresolved symbol _thread_call_free. I'm not calling that (and don't know who is) but as it looked "mach"-ish I added this to OSBundleLibraries: com.apple.kpi.mach 7.0.0 and unresolved symbol went away. But then kld failed with a duplicate symbol: .../PlugIns/IOKit.kext/IOKit definition of absolute _IOGetTime (value 0x7913c) .../PlugIns/Mach.kext/Mach definition of absolute _IOGetTime (value 0x7913c) (BTW: Why is it whining if the value of both definitions is the same?) Does anyone know what's up? Is the whole KPI thing is not baked yet? I've tried to find documentation on it without luck (which probably answers the question). I guess I'll just back out of the KPI stubs but why are they there if they aren't to be used? Any thoughts would be appreciated. Thanks, -Mike _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: Programmatic Back Trace
by Andrew Gallatin 04 Sep '03

04 Sep '03

Sean McBride writes: > > It sure would be nice if the OS provided something like that 'for free' > at kernel level :) It does, but its just not enabled by default. Look into ddb. I think Darwin calls the option "MACH_KDB" BTW, I'd really like to see the shipped kernel emit a symbolic stack trace when it crashes, like FreeBSD 5.x will when the debug.trace_on_panic is set.... Drew _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: Programmatic Back Trace
by Sean McBride 04 Sep '03

04 Sep '03

Geoffrey Schmit (geoffrey.schmit(a)ni.com) wrote on Wed, 3 Sep 2003 11:30:54 -0500: >I'm primarily interested in getting a kernel backtrace from my kernel >extension. I'm porting a library that provides various services to >other kernel extensions we develop. One of these services is to capture >a backtrace. For example, we may capture a backtrace when a kernel >extension detects an error in order to facilitate the debugging of that >error condition. As another example, we may capture a backtrace when a >kernel extension acquires a resource in order to later determine who >acquired that resource if the resource is leaked. Sounds like you want something like the MallocStackLogging env var (and his friends), but at kernel level. Maybe looking at the code for that would be insightful, though I'm sure it wouldn't work without change at kernel level. It sure would be nice if the OS provided something like that 'for free' at kernel level :) -- ____________________________________________________________ Sean McBride, B. Eng Sean.McBride(a)matrox.com Mac Software Designer +1-514-822-6000 Matrox Electronic Systems Ltd. Montrial, Quibec, Canada _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0