November 2010 - launchd-dev - lists.macosforge.org

AXObserverAddNotification error on OS X 10.5 from an agent. What am I doing wrong?
by eveningnick eveningnick 22 Nov '10

22 Nov '10

Hello Why on OS X 10.5 i can't execute AXObserverAddNotification() from an agent, while it works great being called from a process started using a Terminal? (On OX X 10.6 it works great no matter how i am starting my process - whether it's an agent or a "normal" app)? How could i make it work on Mac OS Leopard as agent as well? I have written a simple application that illustrates the problem i'd run into (below), because no one answered to a "dry theoretical question" i asked few times before :) Here is a code, that reproduces the problem: (the function "pid_t find_process(char *process_name)" returns PID of a process by its name, a file "process.c", contains the implementation of it, you can find it here: http://pastebin.com/b7sUbG4m ) ------------------------------file: main.c------------------------------ #include <stdio.h> #include <CoreFoundation/CoreFoundation.h> #include <ApplicationServices/ApplicationServices.h> pid_t find_process(char *process_name); void mycallback(AXObserverRef observer, AXUIElementRef element, CFStringRef notification, void *refcon) { printf("Word Application has been activated!\n"); } int main (int argc, const char * argv[]) { pid_t appPid = find_process("Microsoft Word"); if(appPid==0) { printf("process doesn't exist"); return 1; } AXObserverRef observer; AXError err; err = AXObserverCreate(appPid, mycallback, &observer); if(err!= kAXErrorSuccess) { printf("AXObserverCreate error: %i\n", err); return 1; } AXUIElementRef appElement = AXUIElementCreateApplication(appPid); if(appElement==NULL) { printf("AXUIElementCreateApplication error, returned 0"); } err = AXObserverAddNotification(observer, appElement, kAXApplicationActivatedNotification, NULL); if(err!= kAXErrorSuccess) { printf("AXObserverAddNotification error: %i\n", err); return 1; } CFRunLoopAddSource(CFRunLoopGetCurrent(), AXObserverGetRunLoopSource(observer), kCFRunLoopCommonModes); CFRunLoopRun(); //run runloop indefinitely return 0; } ------------------------------end------------------------------ if you compile that (having added frameworks CoreFoundation and ApplicationServices to the project) and launch it from a terminal, everytime you activate Microsoft Word application, you will get a text "Word Application has been activated" in a terminal window now, if you create a textfile com.test.accessibility.agent.plist, with the following content: ------------------------------com.test.accessibility.agent.plist------------------------------ <?xml version="1.0" encoding="UTF-8"?> <!DOCTYPE plist PUBLIC -//Apple Computer//DTD PLIST 1.0//EN http://www.apple.com/DTDs/PropertyList-1.0.dtd > <plist version="1.0"> <dict> <key>Label</key> <string>com.test.accessibility.agent</string> <key>Program</key> <string>[HERE SPECIFY AN ABSOLUTE PATH TO THE EXECUTABLE]</string> <key>KeepAlive</key> <true/> </dict> </plist> ---------------------------------------------end------------------------------ and then, execute the command from a terminal to install and launch the agent % launchctl load com.test.accessibility.agent.plist % launchctl start com.test.accessibility.agent in system Console (/Applications/Utilities/Console.app) you will see the following output -----------------output in console--------------- com.test.accessibility.agent[1796] AXObserverAddNotification error: -25204 com.apple.launchd[84] (com.test.accessibility.agent[1799]) Exited with exit code: 1 com.apple.launchd[84] (com.test.accessibility.agent) Throttling respawn: Will start in 10 seconds ---------------end--------------- repeating every 10 secs, because of autorelaunch enabled for this agent in plist (don't forget to "%launchctl remove com.test.accessibility.agent" to stop it and delete its plist entry in the end) What causes that? Can i somehow make it work on Mac OS 10.5 as an agent? As i have mnentioned, on OS X 10.6 this is "fixed" (or, rather, changed). Maybe my process, being started by launchd, doesn't have sufficient rights to call this AXObserverAddNotification() function? How can i increase these privileges, and to what level should i increase them? Unfortunately this document http://developer.apple.com/library/mac/#documentation/Accessibility/Referen… doesn't give any information Thank you George

2 1

Re: [launchd-dev] launchd-dev Digest, Vol 41, Issue 5
by Richard Brown 22 Nov '10

22 Nov '10

Per, The method I use is to get the version number of the installed tool and compare it against the version in the app bundle. Since we know that SMJobBless installs the helper executable in /Library/PrivilegedHelperTools, you can create the URL to /Library/PrivilegedHelperTools/com.yourdomain.yourapp.helper and use NSBundle's + bundleWithURL: method and then call - infoDictionary (or similarly use CFBundleCopyInfoDictionaryForURL) to get the helper tool's info dict. You can then query the dictionary for its version, and decide whether to prompt for blessing rights. HTH, Richard On Mon, Nov 15, 2010 at 10:00, <launchd-dev-request(a)lists.macosforge.org> wrote: > From: Per Ejeklint <ejeklint(a)me.com> > To: launchd-dev(a)lists.macosforge.org > Date: Mon, 15 Nov 2010 11:21:07 +0100 > Subject: [launchd-dev] Not calling SMJobBless if not needed? > This is slightly off-topic I guess but the expertise is definitely on this list. :) > > I'm using SMJobBless to install and update a launchd daemon from my app and it works like a charm. Only thing is that the user must authenticate at every launch of the app and not just when the provided daemon is actually newer than the running one. Is there a way work around this with ServiceManagement framework? Or another simple way? I'm new to this part of Mac OS X stuff, so I apologize if I'm posting to the wrong list. > > /Per

3 2

Not calling SMJobBless if not needed?
by Per Ejeklint 15 Nov '10

15 Nov '10

This is slightly off-topic I guess but the expertise is definitely on this list. :) I'm using SMJobBless to install and update a launchd daemon from my app and it works like a charm. Only thing is that the user must authenticate at every launch of the app and not just when the provided daemon is actually newer than the running one. Is there a way work around this with ServiceManagement framework? Or another simple way? I'm new to this part of Mac OS X stuff, so I apologize if I'm posting to the wrong list. /Per

1 0

What would cause `diskimages-helper` to throw an exception in script run by cron, but not launchd?
by Morgan Aldridge 11 Nov '10

11 Nov '10

I have my solution, so I'm mainly just curious as to what the cause of this is and it may be slightly off topic, but this seems the most appropriate place to ask. Basically, I've been using an expect script to run `slapconfig -backupdb` to perform Open Directory backups resulting in encrypted disk images. It worked like a charm on numerous Mac OS X 10.5 Leopard Server installs, run by cron. However, after upgrading each of these servers to Snow Leopard Server, we found the disk images were not created. cron was running the script, but `diskimages-helper` would throw an exception, as seen in /var/log/system.log: Jul 23 03:00:08 servername hdiejectd[93114]: running Jul 23 03:00:11 servername diskimages-helper[93111]: -frameworkCallbackWithDictionary: threw exception calling home. Jul 23 03:00:11 servername diskimages-helper[93111]: threw exception reporting results back to framework. Jul 23 03:00:21 servername hdiejectd[93114]: quitCheck: calling exit(0) The expect script works perfectly fine when run from the shell as root. I wrote a new bash script which uses `serveradmin` (so the wrapper for `slapconfig -backupdb`) and therefore doesn't require expect, but it had the same result. However, as soon as I switched to to being run by launchd instead of cron, it works perfectly. So, what might've changed between Leopard Server and Snow Leopard Server that would cause `diskimages-helper` to throw the exception when run from cron, but _not_ when run by launchd? Morgan Aldridge --- http://www.makkintosshu.com/

1 0

Giving a process privileges, started by an agent, same as to the process, started by Terminal
by eveningnick eveningnick 10 Nov '10

10 Nov '10

Hello! Is it possible? I have an agent, that starts (forks and execs) an another process (when a certain event occurs). But this process started has limited capabilities (in contrary to an application, launched by a user from a terminal). Can i somehow ask a system to grant privileges to make my new spawned by an Agent process to have asme credentials as the one launched "by hands"? (MacOS X 10.5 Leopard)

1 0

Programmatic interface to launchctl and some other questions, OS X 10.5
by eveningnick eveningnick 04 Nov '10

04 Nov '10

Hello I am writing a Cocoa application, that installs and launches an agent (and, if needed - stops that agent, and uninstalls it). This agent is supposed to watch for a launch of another certain application, and if it encounters such an event, it launches some process itself. I am wondering, how could i programmatically register (launchctl's load operation), start (launchctl's start), stop and unload the agent, doing these operations from my Cocoa application? The target operating systems are Mac OS X 10.5 and 10.6. Having searched through the Internet i have found ServiceManagement.h (that allows, as i understand, installing application into /Library/LaunchAgents, even if the application that called SMJobBless/SMJobSumbin was non-root). I am not very "familiar" with OS X ideas. As i understand, when a user doubleclicks my Cocoa application's bundle, the application is launched with non-root privileges, which means i can't "launchctl load" my agent into /Library/LaunchAgents - i can only do that with /Users/username/Library/LaunchAgents directory. But unfortunately this ServiceManagement API is supported only since OS X 10.6 Also there's a /usr/include/launch.h file, which provides some kind of API, but i've no idea how to use it, and i am not sure if it gives any means to load/start/stop/unload agents. Maybe i could use it somehow? Most suggestions about programmatical management of agents are about calling "[NSTask launchProcess:@"launchctl {operation} {label}"]". If there is no other better choice, i'd take it, but i am a little confused about privileges - i won't be able to install an agent to /Library/LaunchAgents, will I?. I'm sorry if i'm saying nonsense, but if i call "[NSTask launchProcess:@"sudo launchctl {operation} {label}"]" will it count as a launch from root, or this trick works only if its done in a terminal? how can i determine, if my agent is running or not, from my Cocoa application? and how could i determine if it's registered (loaded, in terms of launchctl) or not? Calling [NSTask launchProcess:@"launchctl list | grep com.mycompany.myagent"] and then parsing the returned string seems wrong to me :/. Is it really the only choice? I was also searching for the info on how to launch a process, that is a console one (Gui-less). Most suggestions are about using launchd instead of fork/exec-ing it. How is it done usually? Should i necessarily create and save that *.plist to the /Users/username/Library/LaunchAgents/ directory and then call that "[NSTask launchProcess:@"sudo launchctl {operation} {label}"]"?, first with "launchctl load" and second with "launchctl start"? OS X 10.6's ServiceManagement.h API seems to support launching of processes without first copying plist to the one of the "agents'" directories, but again it's only OS X 10.6. Sorry for so many questions :) And Thanks a lot for the response!

5 9

Help with launchd sockets chicken and egg problem
by Dave MacLachlan 02 Nov '10

02 Nov '10

Hey all... I have my master process. It wants to spawn a child process that will startup automatically on reboot. I would like these processes to be able to communicate via sockets, but I don't want to hardcode a port, I would like to use "SecureSocketWithKey" and let launchd take care of starting my process on demand (and clean up the file system as appropriate). Master process puts together a plist and installs it into LaunchAgents. It then calls launchctl and has it load it. The child process calls checks in and is able to get it's socket. How does the master process get the socket on it's end? I understand that if I reboot the system that I can get the socket path via an environment variable, but how do I get it without rebooting? Any info would be greatly appreciated. Cheers, Dave

2 1

Start a process on behalf of launchd
by eveningnick eveningnick 02 Nov '10

02 Nov '10

Hello What is the correct way to launch a process not fork'ing and exec'ing, using launchd instead? I've been reading a lot about how is it more secure, more reliable, and just better :) Should i necessarily create a *.plist file to a /Library/LaunchAgents/ directory, and then after i launched a process, delete from there? Can it be only launched via NSTask'ing launchctl or maybe there's some kind of API for doing such things? Thanks

2 1