October 2006 - Darwin-kernel

Re: On USB bulk device removal, IOService recusively being called. Why ?
by John Dalgliesh 17 Oct '06

17 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi Rohit, Hi all, Also, is there any way that my application should not get the messages of device removal ? Hope this helps. {P^/ Hi, I have developed a KEXT for Mac OS X 10.3.9 ppc and 10.4.7 Intel/PPC for my USB bulk devices.Enountering a strange problem and wanted expert advice from your side: My user application communicates very well with our USB bulk devices through the kext developed. If we unplug the device, the user application gets stopped , as designed ! Now I attach my two USB bulk devices to the root HUB. a) I attach my 1st USB bulk device b) Now I attach my 2nd USB bulk device and start communicating with this device through my user application. Everything;s work fine till now!!! Now if I un-plug the 1st USB bulk device, which has no connection with the communication of user application going with the 2nd USB bulk device, the communication gets closed itself ! Ideally, the second application should have run fine since I the second key is already attached. (Also there is no power drop sceenrio) To reproduce, this scenerio, i attached 1st USB device communicating with application1 and next 5 applications communicating to 5 USB bulk devices seperately. Now if I detach the 1st USB device, the communication of other application also closes itself!!!! But this scenerio does not happen if i unplug my 2,3,4,5th device. /*----Code snippet ---*/ class MyKEXT : public IOService { .. }; IOReturn MyKEXT::message( UInt32 type, IOService * provider, void * argument ) bool MyKEXT::didTerminate( IOService * provider, IOOptionBits options, bool * defer ) bool MyKEXT::willTerminate( IOService * provider, IOOptionBits options ) /*---------*/ In didTerminate function, I am closing my provider and freeing the memory I allocated. In willTerminate function, I am aborting the pipes. My USB bulk has a default pipe, IN and OUT pipe. In message function, I am just calling the superclass to the handle the message first. Basically my cleanup functionality is in didTerminate and willTerminate function and not in message function. So when we remove the device, the message function calls the message of IOUserClient. Below is the code snippet for IOUserClient part: /*----Code snippet---*/ IOReturn MyDriverUserClient::message(UInt32 type, IOService * provider, void * argument) { IOReturn retVal = kIOReturnSuccess; switch ( type ) { case kIOMessageServiceIsTerminated : case kIOMessageServiceIsRequestingClose : retVal = terminate (); break; default: break; } return retVal; } bool MyDriverUserClient::finalize(IOOptionBits options) { return super::finalize(options); } bool MyDriverUserClient::terminate(IOOptionBits options) { return super::terminate(options); } /*-----*/ Now when I unplug the first device, the call from IOService message goes for each IOUserclient message that calls the terminate function for each application communicating. Why so ?? Ideally the call should only go the first devices's UserClient. It seems that the call is recusively going to each client that is communicating.Why is it happening ? It seems that internally, IOSservice* is recusively being called and terminate is being called for each user application. What can be the possible issue ? Am i missing something / or doing something wrong ? Or is there might be some issue with the IOKit iself? Did anybody enountered the similar issue before ? Please send your useful suggestions, as to why it is happening and how to avoid this issue. Please feel free to ask if my query is not clear! Thanks and Regards, Rohit Dhamija _______________________________________________ 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.… On Mon, 16 Oct 2006, rohit dhamija wrote: Please let me know in case my asked query is not clear! I think your query is clear. Most likely no-one knows the answer. Maybe darwin-drivers would have been a better list for this IOKit question 'tho. It does sound to me like a bug in either your KEXT or your app, but to verify this you should do some debugging. This is what I'd do: - add some logging to your IOService 'message' override to see if the terminate messages are occuring in the kext. If not then it's probably an app problem. - add an 'identify' function to your UserClient interface to make sure the app is really connected to the device you think it is. It could return e.g. the 'this' pointer and you could compare it against the system log. - OTOH if the terminate messages are in the kernel then do a bit of two-machine debugging (or just panic and look at the backtrace) to find out why it is there... the source for IOService is all available and it's not too hard to follow along with it. I doubt it. But it might be an idea to also set up device removal notifications to make sure instead of relying on message, i.e. IOServiceAddMatchingNotification with kIOTerminatedNotification. Rohit On 10/15/06, rohit dhamija <rohit.dhamija(a)gmail.com> wrote: This email sent to site_archiver(a)lists.apple.com

1 0

Re: On USB bulk device removal, IOService recusively being called. Why ?
by rohit dhamija 16 Oct '06

16 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:to:subject:in-reply-to:mime-version:content-type:references; b=o0b3OkMdp3aeSg/bURb/hZxyELcowtSCRKfkh+PBwxqugbUoTf2fCtC3ktoJFJneXvZQsa4B7fxIX5tk7idEavddZrZl+cVudvH3m+Py5Qu1EK2hG2sZng4kRcSmgWl0O+KL62EGK8ZmElBr2Cp3f1W1HFVgwPxytrXcFdtzzkY= _______________________________________________ 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: IOServiceOpen fails
by plumber Idraulico 16 Oct '06

16 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com one way is a Notification system cf : documentation -plum Hi, I have written a kernel driver.Now want to talk to it through a userclient.My user space application is a cocoa application which calls IOServiceOpen() to establish a connection with the userclient.The call too IOServiceOpen fails and returning a value of hex value of 0xe00002c7 (-536870201 in decimal).But the API's IOMasterPort(),IOServiceMatching() & IOServiceGetMatchingServices() are returning success.Can anybody help me out in this matter. The info.plist file was corrupted. I created a new one with "IOUserClientClass" and its name as key-value pair. Now it works.I am able to send data to the driver from my userspace program :-) .I need to send data back to my user space cocoa app from my driver.How can this be done? Regards, Sanjay _______________________________________________ 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:IOServiceOpen fails
by sanjay yaragatti 16 Oct '06

16 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.co.in; h=Message-ID:Received:Date:From:Subject:To:Cc:MIME-Version:Content-Type:Content-Transfer-Encoding; b=ZHcy0b8Xouj6NLR8ciHkea8qAfWEZigm4S90+tpvyn8yxdMUyHz3cF5f0+ZNbWFjKiwMBJEnRngwaJK+T1Ns3aPUtT5QVw6fn0pyzt5nKDx5qzWqTQpQJGhEY/lJWjzfH5HSefQQnYi37pNSbqpEMfc6s41KLtRVEoK0f7PjwBg= ; Hi, >I have written a kernel driver.Now want to talk to it >through a userclient.My user space application is a >cocoa application which calls IOServiceOpen() to >establish a connection with the userclient.The call >too IOServiceOpen fails and returning a value of hex >value of 0xe00002c7 (-536870201 in decimal).But the >API's IOMasterPort(),IOServiceMatching() & >IOServiceGetMatchingServices() are returning >success.Can anybody help me out in this matter. The info.plist file was corrupted. I created a new one with "IOUserClientClass" and its name as key-value pair. Now it works.I am able to send data to the driver from my userspace program :-) .I need to send data back to my user space cocoa app from my driver.How can this be done? Regards, Sanjay __________________________________________________________ Yahoo! India Answers: Share what you know. Learn something new http://in.answers.yahoo.com/ _______________________________________________ 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: Setting Explicit Socket Permissions!!
by Justin C. Walker 16 Oct '06

16 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Oct 15, 2006, at 20:30 , Ravi Yadav wrote: "man 2 chmod" Justin -- Justin C. Walker, Curmudgeon at Large Institute for the Absorption of Federal Funds ----------- My wife 'n kids 'n dogs are gone, I can't get Jesus on the phone, But Ol' Milwaukee's Best is my best friend. ----------- _______________________________________________ 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.… Hi, Once I have installed my app, Every time its loaded my app creates and listens on a specific socket ie /tmp/MYSOCKET. Every time the app load and creates the socket I want to :- 1)Set some specific permissions on /tmp/MYSOCKET, irrespective of what socket permissions it has been created with by default. How can I achieve this in my app programitically? Note that you have to have permissions to modify directory entries; in particular, if you aren't running as root, this will depend on the owner and group associated to the directory enty. This email sent to site_archiver(a)lists.apple.com

1 0

Setting Explicit Socket Permissions!!
by Ravi Yadav 16 Oct '06

16 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thread-index: Acbw01xUiee4KtP/QT6vz/Thx2rBaA== Thread-topic: Setting Explicit Socket Permissions!! _______________________________________________ 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: System call for OS version info
by plumber Idraulico 16 Oct '06

16 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com it was humour/joke +the UNIX philosophy ( Terry Lambert the new pape ) the the the :D :D it's not THE problem it's not a story of UNIX philosophy ... 1 secure code -> don't link a base tool against x libs this is only view i can share -plum Tony Plus, the UNIX philosophy is to build new tools out of old tools, instead of reinventing the wheel. 8-). Plus Plus your software really should not care about the build number or version anyway... so making it harder to get discourages people writing code for programs to check the version before deciding they are willing to run. _______________________________________________ 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 both agree and disagree with your philosophy Terry. My app is "memtest" and it runs as a unix command-line utility in single-user mode. So yes, I NEED to stay away from frameworks (CF specifically) because they don't yet work in 64-bit mode. Memtest is a 64-bit app on 64-bit machines so I can't include anything that won't run on 64- bit arch's. I'm always looking for kernel-level or other low-level system calls to gather the info I need. I use sysctl extensively and it sure would have been nice if there was a low-level call to get the MacOS version info as well. Gestalt worked great for this purpose under OS 9. So far, I haven't used anything "non-unix" in memtest and I intend to keep it that way. I don't use the system info to decide whether to run or not. It's used only for debugging purposes when a user reports a runtime problem with memtest. See my previous post for an example of one debugging nightmare. Since then, I've been adding more system information to the logfile so I don't have to bug the user for this info if they have a problem. Remember the "dlylib" issue with 10.3.9? I wasn't the only app broken by the missing lib. I recall seeing a bunch of developers having the same problem as me back then. So although in an ideal world we really shouldn't care what build we're running under, in the real world it can make a difference. And when it does, I want all the info I can get about the user's runtime environment. What I'm saying is that a Gestalt- type call that would give all the OS info provided by "sw_vers" would be really nice so XML parsers wouldn't be needed. :-) Terry Lambert <tlambert(a)apple.com> wrote: Well, I'm a kernel person - so anything that avoids linking an XML library or CF Framework to get the same information is a win in my book. -- Terry This email sent to site_archiver(a)lists.apple.com

1 0

Fwd: Re: System call for OS version info
by Tony Scaminaci 16 Oct '06

16 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=sbcglobal.net; h=Message-ID:Received:Date:From:Subject:To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=iUQlXCCP8MPp7g9xxXSIM2uRs0NlicOyKoYQfGWU28TNwOZrzZsQ9TyP02q97nf+X0I7Z4G+rF0xedPnSMQRlRdpc2+bkS/Zsl6kwYN+4S/xCo2CAHaoLRYykxlM6c4NEhG0Bsuk2+OpFP5aLFlQVuHy6p05I2OHBftCZYHiEpw= ; _______________________________________________ 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: System call for OS version info
by Tony Scaminaci 16 Oct '06

16 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=sbcglobal.net; h=Message-ID:Received:Date:From:Subject:To:In-Reply-To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=lp8TKf4FSxsppz9J971l95k4OO9tTreNkEJmEaRnSUihCdGNYreX+YnYY3NAY7kDL3QYM16NSB2pOxTS7Z32zp083XL4qGZ2cnQY+cGJKkZP19WzP45cFkXPjFcDrEV/oZ4kMoRap8io5HkwCRejz6sudAdqoDwbgi9iaQ8o7G8= ; _______________________________________________ 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: System call for OS version info
by Terry Lambert 15 Oct '06

15 Oct '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Oct 15, 2006, at 4:16 PM, plumber Idraulico wrote: uggly -- Terry _______________________________________________ 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.… -plum Well, I'm a kernel person - so anything that avoids linking an XML library or CF Framework to get the same information is a win in my book. Plus, the UNIX philosophy is to build new tools out of old tools, instead of reinventing the wheel. 8-). Plus Plus your software really should not care about the build number or version anyway... so making it harder to get discourages people writing code for programs to check the version before deciding they are willing to run. This email sent to site_archiver(a)lists.apple.com

1 0