January 2004 - Darwin-kernel

Re: scheduling question
by rimas avizienis 16 Jan '04

16 Jan '04

i have written a low latency coreaudio driver which sends and receives UDP packets containing audio out over the network interface to a special purpose piece of hardware i've built. In order to get the performance/latency/etc i needed when the machine was running under a heavy load (doing real time processing of live audio, etc) I had to modify the AppleGMACEthernet.kext driver so that the main loop in my coreaudio driver (get packet, copy buffers, call taketimestamp, send a packet out) gets called at primary interrupt time (highest priority in the system) by using checkForInterrupt instead of just interruptOccurred and having it call my coreaudio driver. I basically bypassed the whole networking stack. You might need to do something of that sort... -rimas --- Mike Smith <drivers(a)mu.org> wrote: > On Jan 15, 2004, at 1:31 AM, Stefan Werner wrote: > > > Are there other tweaks to enhance the latency of > networking? I don't > > mind a decrease in throughput, I'm in one of the > rare cases where > > bandwidth is not the limit. There are these > popular "bandwidth tuner" > > tweaks where people change the system's TCP buffer > size with sysctls, > > would shrinking the buffers give me any latency > improvements or will > > that result in dropped packets? > > It won't help; the system already tries to give you > the lowest possible > latency. There's not really any more blood to > squeeze from the stone. > > It's important, back in the design stage, to take > into consideration > that you're going to be running in a multiprogrammed > environment and > try to take this into account there, rather than > trying to tweak a > system that can't give you guarantees to fit a > design that assumes a > particular value of "reasonable". > > Obviously this isn't always possible. > > = 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. __________________________________ Do you Yahoo!? Yahoo! Hotjobs: Enter the "Signing Bonus" Sweepstakes http://hotjobs.sweepstakes.yahoo.com/signingbonus _______________________________________________ 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

Processor Temp
by Robert J. Accettura 16 Jan '04

16 Jan '04

Is there a way to get the processor temp via terminal in OS X? I'd like to script some method to record the processor temp of my old beige G3. -- Robert J. Accettura robert(a)accettura.com [demime 0.98b removed an attachment of type application/x-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

DiskArbitration API Changes for MacOSX 10.3.2?
by Lance Drake 15 Jan '04

15 Jan '04

Hi Darwin folks, My process now dies as shown below: Exception: EXC_BAD_ACCESS (0x0001) Codes: KERN_PROTECTION_FAILURE (0x0002) at 0x00000000 Thread 0 Crashed: 0 com.apple.CoreFoundation 0x901aab90 CFSetRemoveValue + 0x38 1 com.apple.DiskArbitration 0x907f5294 __DiskArbDiskUnmountCallback + 0x170 2 com.apple.DiskArbitration 0x907f1fe8 _DADispatchCallback + 0x108 3 com.apple.DiskArbitration 0x907f1374 _DASessionCallback + 0x114 4 com.apple.CoreFoundation 0x901ad4d0 __CFMachPortPerform + 0xe0 5 com.apple.CoreFoundation 0x901a9c54 __CFRunLoopDoSource1 + 0xc8 6 com.apple.CoreFoundation 0x90191b20 __CFRunLoopRun + 0x540 7 com.apple.CoreFoundation 0x901960bc CFRunLoopRunSpecific + 0x148 8 com.apple.CoreFoundation 0x901ff514 CFRunLoopRun + 0x34 9 sfpatd 0x00014488 SetupAndRunDriver + 0x1c4 10 sfpatd 0x000141e4 main + 0x2a0 This 100% repeatable event occurs after I request the DA kext to unmount a volume. There is no such problem when the same software is run under MacOSX 10.2.8. Can anyone shed any light as to how best to ameliorate this situation? Should I file a bug for this behavior - or is there some good reason why it's OK? Thanks very much, Lance Drake _______________________________________________ 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: TAKE2: Problems testing with IOStorageFamily.kext
by Chris Walters 15 Jan '04

15 Jan '04

Dan Markarian wrote: Hi Chris, It sounds like your open on the device was closed at some point. The strategy will return an EACCES on a read should that device not still be open. The Storage Family does not return EACCES on a read in any other context, however, it is plausible the error came from a driver below the family, of course. This is the case... the device was closed, but I'm pretty sure it wasn't by our driver. If I load my FileSystem from within a login the strategy succeeds. What I find is when I log out and then back in (even as the same user) I get the EACCES error. I initially ran into this problem by starting my FileSystem on boot and trying to do tests as a user. I have traced using gdb into the media layer which returns the error. I can only assume that another entity is closing the device (maybe prompted by the logout or login). This is why I want to have a debug version of IOStorageFamily so I can see when the device is closed. Any ideas? Thanks. Dan On Jan 14, 2004, at 6:33 PM, Chris Walters wrote: I'm developing a File System which is getting an EACCES error from doing a strategy call to the IOStorageFamily. I need to be able to build and install a debug version of IOStorageFamily. Building it is easy (with either ProjectBuilder or pbxbuild)... the problem is when I install and restart the system I get the "No Entry" symbol. Could somebody explain why my kext isn't working... * do I have to copy more than just the kext? * is there a difference between building with ProjectBuilder or pbxbuild? Thanks. _______________________________________________ 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. _______________________________________________ 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. _______________________________________________ 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: IOATAController device blocking bus
by RJ Auburn 15 Jan '04

15 Jan '04

*This message was transferred with a trial version of CommuniGate(tm) Pro* Garth, Thanks for the info... It also looks like IOATADevConfig.cpp is missing the closing brace at the end of the file. Once I fixed that it all complies ok... RJ --- RJ Auburn CTO, Voxeo Corporation tel:+1-407-418-1800 On Jan 15, 2004, at 10:41, Garth Cummings wrote: *This message was transferred with a trial version of CommuniGate(tm) Pro* Hi RJ, On Jan 14, 2004, at 10:21 AM, RJ Auburn wrote: *This message was transferred with a trial version of CommuniGate(tm) Pro* Is anyone else having problems getting IOATAFamily to compile? When try to compile the 10.3.2. version I get the following errors that are below. Maybe I'm missing some magic bit of the build setup or something. RJ <snippage> In file included from IOATAController.h:34, from ATADeviceNub.cpp:36: ATATimerEventSource.h:34:1: unterminated #ifndef from ATADeviceNub.cpp:36: In file included from IOATAController.h:34, ATATimerEventSource.h:34:1: unterminated #ifndef No, the compiler message is correct. Somehow the version of ATATimerEventSource.h in Darwin lost the last line, which should read: #endif /*_ATATIMEREVENTSOURCE_H*/ I've passed this on to the folks internally to figure out how this happened. Thanks, --gc _______________________________________________ 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: scheduling question
by Jim Magee 15 Jan '04

15 Jan '04

On Jan 15, 2004, at 4:31 AM, Stefan Werner wrote: So, using THREAD_PRECEDENCE_POLICY will only increase the priority of the thread within my process, but the priority of my thread will not be affected? If so, increasing the process' priority in addition should then give me the desired system-wide effect, I assume. Yes, but unless you wrote every application/thread in the system, you'd only be guessing at a priority for your task (and potentially starting an arms race with all the other developers towards the highest possible priority - and then you're all just back to a simple round-robin scheme once you get there). That's why we recommend that the application developer NOT tweak the task/process-wide priority values. For one thing, if it's a GUI application, this will happen automatically as you switch in and out out of "front-most" application status. It wouldn't be good to have multiple entities fighting over those task-wide priority values in these cases. One more question: If I change a thread's scheduling policy, will threads created from that one inherit that policy or will they have normal policy? The creating thread is irrelevant. New threads pick up the default scheduling attributes of their containing task - or the supplied attributes on the create call (for pthread_create()). --Jim _______________________________________________ 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: IOATAController device blocking bus
by Garth Cummings 15 Jan '04

15 Jan '04

Hi RJ, On Jan 14, 2004, at 10:21 AM, RJ Auburn wrote: *This message was transferred with a trial version of CommuniGate(tm) Pro* Is anyone else having problems getting IOATAFamily to compile? When try to compile the 10.3.2. version I get the following errors that are below. Maybe I'm missing some magic bit of the build setup or something. RJ <snippage> In file included from IOATAController.h:34, from ATADeviceNub.cpp:36: ATATimerEventSource.h:34:1: unterminated #ifndef from ATADeviceNub.cpp:36: In file included from IOATAController.h:34, ATATimerEventSource.h:34:1: unterminated #ifndef No, the compiler message is correct. Somehow the version of ATATimerEventSource.h in Darwin lost the last line, which should read: #endif /*_ATATIMEREVENTSOURCE_H*/ I've passed this on to the folks internally to figure out how this happened. Thanks, --gc _______________________________________________ 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: TAKE2: Problems testing with IOStorageFamily.kext
by Dan Markarian 15 Jan '04

15 Jan '04

Hi Chris, It sounds like your open on the device was closed at some point. The strategy will return an EACCES on a read should that device not still be open. The Storage Family does not return EACCES on a read in any other context, however, it is plausible the error came from a driver below the family, of course. Dan On Jan 14, 2004, at 6:33 PM, Chris Walters wrote: I'm developing a File System which is getting an EACCES error from doing a strategy call to the IOStorageFamily. I need to be able to build and install a debug version of IOStorageFamily. Building it is easy (with either ProjectBuilder or pbxbuild)... the problem is when I install and restart the system I get the "No Entry" symbol. Could somebody explain why my kext isn't working... * do I have to copy more than just the kext? * is there a difference between building with ProjectBuilder or pbxbuild? Thanks. _______________________________________________ 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. _______________________________________________ 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: scheduling question
by Mike Smith 15 Jan '04

15 Jan '04

On Jan 15, 2004, at 1:31 AM, Stefan Werner wrote: Are there other tweaks to enhance the latency of networking? I don't mind a decrease in throughput, I'm in one of the rare cases where bandwidth is not the limit. There are these popular "bandwidth tuner" tweaks where people change the system's TCP buffer size with sysctls, would shrinking the buffers give me any latency improvements or will that result in dropped packets? It won't help; the system already tries to give you the lowest possible latency. There's not really any more blood to squeeze from the stone. It's important, back in the design stage, to take into consideration that you're going to be running in a multiprogrammed environment and try to take this into account there, rather than trying to tweak a system that can't give you guarantees to fit a design that assumes a particular value of "reasonable". Obviously this isn't always possible. = 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: scheduling question
by Stefan Werner 15 Jan '04

15 Jan '04

Hi, On Wednesday, Jan 14, 2004, at 22:05 Europe/Berlin, Jim Magee wrote: There are "helper" threads that process the data within the in-kernel network code somewhat isolating your thread from directly accessing the network hardware in that way. Your realtime thread may even interfere with the responsiveness of these helper threads in ways that might affect your own data. Even if you don't interfere with the processing of the in-kernel data, your response-time characteristics are going to be dominated by the scheduling behavior of these in-kernel threads. There are also situation where the network stack just borrows another user-level thread (that just happened to be in the network stack at the time) to drive completion of your packet handling (inbound and outbound). Obviously, you can't make these all realtime threads, so that too will affect your latency. Are there other tweaks to enhance the latency of networking? I don't mind a decrease in throughput, I'm in one of the rare cases where bandwidth is not the limit. There are these popular "bandwidth tuner" tweaks where people change the system's TCP buffer size with sysctls, would shrinking the buffers give me any latency improvements or will that result in dropped packets? The THREAD_PRECEDENCE_POLICY isn't really a completely different policy from the normal timeshare (migrating priority) policy on the system. It's just a way to control a key aspect of that policy. In essence, you are telling the schedule that, within this one process, some threads should have precedence over others even if they have the same priority. Essentially, you are assigning ordinal values to the process' threads without having to know or guess an appropriate system-wide priority for the threads. The system doesn't guarantee to honor it, but will attempt to do so (fairness may dictate that we chose threads to execute out of ordinal order). However, the effect should be process-local, and therefore cannot be relied upon to have any system-wide effect whatsoever. So, using THREAD_PRECEDENCE_POLICY will only increase the priority of the thread within my process, but the priority of my thread will not be affected? If so, increasing the process' priority in addition should then give me the desired system-wide effect, I assume. What has worked well for others in this type of situation is to use the THREAD_EXTENDED_POLICY notion to declare the thread as a "non-long-running" thread (by specifying the timeshare parameter as FALSE). As long as your thread is well behaved and gives up the CPU in reasonable amounts of time, it will always run at a fixed priority. I think that's what one of the code snippets I tried is doing. I got that from someone on the CoreAudio list when I asked for low-latency scheduling tips. One more question: If I change a thread's scheduling policy, will threads created from that one inherit that policy or will they have normal policy? As such, it will have significant preference over other threads in the system that are (or have been) consuming large amounts of CPU time (and are therefore likely to get in the time way of reasonable timeliness). The main competitor for my network thread is a CoreAudio thread with a number of processing-intensive AudioUnits. Well, it could be that some of the problems I have now on my development iBook disappear by themselves when we run it on the dual G5 it's intended to run on. -Stefan _______________________________________________ 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