January 2004 - Darwin-kernel

Re: TAKE2: Problems testing with IOStorageFamily.kext
by Chris Walters 19 Jan '04

19 Jan '04

Dan Markarian wrote: Hi Chris, You could just insert your own d_open and d_close routines which track what you need and call into the original function when done. I've traced the use of these function calls and my filesystem does not close the device... it appears that something else behind the scenes on a logout/login does it. If the device is reopened it works happily... but I would expect that a user context shouldn't influence a devices state. I am not sure why the debug driver does not load. You can verify it with the following command after installing it. Be certain that the source it built on the appropriate system version as well. kextload -nt /System/Library/Extensions/IOStorageFamily.kext I ran this and it said the kext appears to be valid. If I reboot with the partition as the default startup system it displays an apple with a black line above it (appearing like it is displaying text, maybe the reason why it fails, but the box isn't big enough). If I boot with the partition not as default then a "no entry" sign appears. Note: I failed to mention the OS is 10.2.8 on G4. Dan On Jan 15, 2004, at 5:59 PM, Chris Walters wrote: 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. _______________________________________________ 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: Processor Temp
by Robert J. Accettura 18 Jan '04

18 Jan '04

Well, I found some info here: http://e-www.motorola.com/files/32bit/doc/app_note/AN1800.pdf But I'm missing something :-( I guess. Most likely licensing is in the way. Perhaps someone knows better. Also found this: http://mail-index.netbsd.org/port-powerpc/2002/03/03/0000.html But nothing more to it. I'm sure it's possible. Under Mac OS 9, many software products did it, so there is a way to interface to the hardware and retrieve the info. The question is how under OS X? More interesting is it apparantly has a PPC 750: http://www.bresink.de/osx/TemperatureMonitor.html Seems supported. Shawn Erickson wrote: > > > On Jan 15, 2004, at 8:11 PM, Robert J. Accettura wrote: > > 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. > > > Note the following email... it may give a avenue but as other have > stated no defined interface exists. > > Begin forwarded message: > > *From: *Shawn Erickson <shawn(a)freetimesw.com> > *Date: *January 8, 2004 8:06:34 AM PST > *To: *Kory T <talmage(a)zero.ad.jp> > *Cc: *cocoa-dev(a)lists.apple.com > *Subject: Re: Sensor readings > * > On Jan 7, 2004, at 11:25 PM, Kory T wrote: > > Hi, > > I've been trying to figure out where the readings for the G5 > temp sensors can be located (also for Xserves). I'm assuming > OS X has this info in a way where I can just 'cat' it (since > everything on nix is treated as a file). Or is there a > framework that I'm not aware of that allows me to access this > info (still new to programming on OS X ;)? Can someone point > me to the right direction? > > > I hopefully can get on the path at least... (don't know that > answer myself at the moment) > > Try running the following command: "ioreg -b -n IOHWSensor" > > You will see entries like the following... > > | | | | +-o temp-sensor <class IOService, registered, matched, > active, busy 0, retain count 6> > | | | | +-o IOHWSensor <class IOHWSensor, registered, matched, > active, busy 0, retain count 6> > | | | | { > | | | | "current-value" = 1847296 > | | | | "polling-period" = 18446744073709551615 > | | | | "location" = "DRIVE BAY" > | | | | "type" = "temperature" > | | | | "IOClass" = "IOHWSensor" > | | | | "IONameMatched" = "temp-sensor" > | | | | "version" = 1 > | | | | "zone" = <00000001> > | | | | "IOProbeScore" = 0 > | | | | "IOMatchCategory" = "IODefaultMatchCategory" > | | | | "IONameMatch" = > ("temp-sensor","voltage-sensor","current-sensor","power-sensor","adc-sensor","fanspeed-sensor") > > | | | | "CFBundleIdentifier" = "com.apple.driver.AppleHWSensor" > | | | | "IOProviderClass" = "IOService" > | | | | "sensor-id" = 4 > | | | | } > > Using IOKit methods you can scan the IO registry programatically > to find the above information (see IOKit / Kernel document on > Apples developer site). You can also get a hold of the driver > instances (instances of IOHWSensor) which may or may not provide a > user client that will allow you directly communication with > whatever API it exposes. I don't know the units or encoding for > current-value... > > Also sysclt may exist for the above information... (I don't know > myself) > > -Shawn > > > -- 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

Re: Processor Temp
by Shawn Erickson 18 Jan '04

18 Jan '04

On Jan 15, 2004, at 8:11 PM, Robert J. Accettura wrote: > 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. Note the following email... it may give a avenue but as other have stated no defined interface exists. Begin forwarded message: > From: Shawn Erickson <shawn(a)freetimesw.com> > Date: January 8, 2004 8:06:34 AM PST > To: Kory T <talmage(a)zero.ad.jp> > Cc: cocoa-dev(a)lists.apple.com > Subject: Re: Sensor readings > > On Jan 7, 2004, at 11:25 PM, Kory T wrote: > >> Hi, >> >> I've been trying to figure out where the readings for the G5 temp >> sensors can be located (also for Xserves). I'm assuming OS X has >> this info in a way where I can just 'cat' it (since everything on nix >> is treated as a file). Or is there a framework that I'm not aware of >> that allows me to access this info (still new to programming on OS X >> ;)? Can someone point me to the right direction? > > I hopefully can get on the path at least... (don't know that answer > myself at the moment) > > Try running the following command: "ioreg -b -n IOHWSensor" > > You will see entries like the following... > > | | | | +-o temp-sensor <class IOService, registered, > matched, active, busy 0, retain count 6> > | | | | +-o IOHWSensor <class IOHWSensor, registered, > matched, active, busy 0, retain count 6> > | | | | { > | | | | "current-value" = 1847296 > | | | | "polling-period" = 18446744073709551615 > | | | | "location" = "DRIVE BAY" > | | | | "type" = "temperature" > | | | | "IOClass" = "IOHWSensor" > | | | | "IONameMatched" = "temp-sensor" > | | | | "version" = 1 > | | | | "zone" = <00000001> > | | | | "IOProbeScore" = 0 > | | | | "IOMatchCategory" = "IODefaultMatchCategory" > | | | | "IONameMatch" = > ("temp-sensor","voltage-sensor","current-sensor","power-sensor","adc- > sensor","fanspeed-sensor") > | | | | "CFBundleIdentifier" = > "com.apple.driver.AppleHWSensor" > | | | | "IOProviderClass" = "IOService" > | | | | "sensor-id" = 4 > | | | | } > > Using IOKit methods you can scan the IO registry programatically to > find the above information (see IOKit / Kernel document on Apples > developer site). You can also get a hold of the driver instances > (instances of IOHWSensor) which may or may not provide a user client > that will allow you directly communication with whatever API it > exposes. I don't know the units or encoding for current-value... > > Also sysclt may exist for the above information... (I don't know > myself) > > -Shawn _______________________________________________ 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 18 Jan '04

18 Jan '04

Hi, > Message: 1 > Date: Thu, 15 Jan 2004 22:19:40 -0800 (PST) > From: rimas avizienis <funkrhythm(a)yahoo.com> > > 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... That woud be more or less be exactly what I need, as I am also working on an audio over Ethernet application, except that I have a PowerMac on both ends. While I'd love to stay in userland wherever possible and don't have any plans to move the receiving side to the kernel, I'd be interested in any tweaks you did to the audio driver. My current audio driver .kext is using IOKit timers to replace the hardware interrupt, and I am experiencing problems when going below 128 frames @ 48kHz. I am in the lucky situation where bandwidth is not the problem (switched GBit), but if there are ways to get a lower latency I'd love to hear about them. -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

Re: scheduling question
by Stefan Werner 17 Jan '04

17 Jan '04

Hi! On 15.01.2004, at 22:29, Jim Magee wrote: 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. I don't intend to tweak it in the application itself but was thinking of using renice. It's more of a server-type application that doesn't involve user interaction, and in most cases it'll be the only application running on that computer (except for system daemons). Thanks for your input, you really helped me. -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

Re: Processor Temp
by Mike Smith 17 Jan '04

17 Jan '04

On Jan 16, 2004, at 8:36 PM, Robert J. Accettura wrote: Thanks for the information. I hoped it was possible, as several Apps did so under OS 9 with success (SonnetTech's Metronome, XLR8's Mach Speed Control I believe did the same... Newer Tech had a utility as well). I was hoping there was some way to achieve the capturing of that data with minimal work. It's less useful than you think. Without a good idea of what constitutes a good vs. not good temperature, it's a mostly-useless number. However, the bottom line is that whilst in some cases you can get something resembling a temperature reading from some systems, there's no standard interface that you can use. = 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: Processor Temp
by Robert J. Accettura 17 Jan '04

17 Jan '04

Thanks for the information. I hoped it was possible, as several Apps did so under OS 9 with success (SonnetTech's Metronome, XLR8's Mach Speed Control I believe did the same... Newer Tech had a utility as well). I was hoping there was some way to achieve the capturing of that data with minimal work. Sam Hart wrote: > Robert wrote: > >> Is there a way to get the processor temp via terminal in OS X? > > > I doubt it > > There used to be a function called something like > getCoreProcessorTemperature() in pre-OSX, > however I think it has been deprecated ( it was commented out in the > header ). > > The reason it doesn't appear ( if my memory serves me correct ) is > that not all PPC support > manual calibration of the TAU, I can only assume that the chip > handles it, but I could be wrong. > >> I'd like to script some method to record the processor temp of my old >> beige G3. > > > Well I had it working on a G3 and all my TAU documentation refers to > the G3 ( 750L I think ). > > If you can avoid doing it I would, the words "DANGER DANGER POISON" > should be written all over it. > > You would have to do it in kernel space as you would have to be in > supervisor mode. > > It is not as simple as saying to the chip "How hot are you?" you set > a pair of interrupts > ( if I remember correctly, which I probably don't ) and pick a low > value, hoping to be interrupted > straight away, you increase the value until it no longer interrupts, > and then you should have your value. > > There is a kext in Jaguar called "AppleCPUThermo.kext", however I > don't know what it does. > > I write all this so that someone might put me straight or fill in the > gaps. > > There are also a couple of apps that might help > > <http://homepage.mac.com/takashi_hamada/Acti/MacOSX/ThermoInDock/ > index.html> > > <http://www.kezer.net/thermographx.html> > > though I do not know if they support G3, or log to a file. > > > S a m > _______________________________________________ > 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. > -- 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

Re: DiskArbitration API Changes for MacOSX 10.3.2?
by Dan Markarian 16 Jan '04

16 Jan '04

Hi Lance, I am not sure how it could crash in that context. You might wish to confirm that it is not a latent memory smasher in your process. If you find no leads, go ahead and file a radar. I know of no similar problem currently reported. Dan On Jan 15, 2004, at 6:49 PM, Lance Drake wrote: 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. _______________________________________________ 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 16 Jan '04

16 Jan '04

Hi Chris, You could just insert your own d_open and d_close routines which track what you need and call into the original function when done. I am not sure why the debug driver does not load. You can verify it with the following command after installing it. Be certain that the source it built on the appropriate system version as well. kextload -nt /System/Library/Extensions/IOStorageFamily.kext Dan On Jan 15, 2004, at 5:59 PM, Chris Walters wrote: > 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: Processor Temp
by Sam Hart 16 Jan '04

16 Jan '04

Robert wrote: Is there a way to get the processor temp via terminal in OS X? I doubt it There used to be a function called something like getCoreProcessorTemperature() in pre-OSX, however I think it has been deprecated ( it was commented out in the header ). The reason it doesn't appear ( if my memory serves me correct ) is that not all PPC support manual calibration of the TAU, I can only assume that the chip handles it, but I could be wrong. I'd like to script some method to record the processor temp of my old beige G3. Well I had it working on a G3 and all my TAU documentation refers to the G3 ( 750L I think ). If you can avoid doing it I would, the words "DANGER DANGER POISON" should be written all over it. You would have to do it in kernel space as you would have to be in supervisor mode. It is not as simple as saying to the chip "How hot are you?" you set a pair of interrupts ( if I remember correctly, which I probably don't ) and pick a low value, hoping to be interrupted straight away, you increase the value until it no longer interrupts, and then you should have your value. There is a kext in Jaguar called "AppleCPUThermo.kext", however I don't know what it does. I write all this so that someone might put me straight or fill in the gaps. There are also a couple of apps that might help <http://homepage.mac.com/takashi_hamada/Acti/MacOSX/ThermoInDock/ index.html> <http://www.kezer.net/thermographx.html> though I do not know if they support G3, or log to a file. S a m _______________________________________________ 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