September 2007 - Darwin-dev

Re: Thread Priority?
by Basavaraj 21 Sep '07

21 Sep '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:user-agent:mime-version:to:cc:subject:references:in-reply-to:content-type:content-transfer-encoding; bh=az5gi9uRC7SGKgV5PM+Wzcp09FACKd21bq7rlxX1Y5w=; b=hwiMy+gwbamlVRInGJUrNzRSV6fRszNuxGhUsBteQdSieW+QkRMIiPpuHDYX2hJYDG8L3sTipv3/j4zT2pvG2OcNgJgj0UcnCcJNLiGXUHqjJ7zcHUBHw6GZe9D8GJ9iWRCspLskAG2ZTeoSU+J6dHvwlo1sK0dbi9G2CuFKqPM= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:user-agent:mime-version:to:cc:subject:references:in-reply-to:content-type:content-transfer-encoding; b=j5KKUvL71xcbNfrn8EePfhWrzJ3QKedMIaqgyQo6rdPsWDhQRzMdUvv++Me+BCaFiAV0k3OVjoUc+Iie1O4FH2r1gTYB34VvR7OuuIQMmAfA2Cl5utoJMz6u0b9+1HrFvK4Yw5gCKgsJ1mVRI1szF9u8HnDL/dUlqHyrUVtWVxU= User-agent: Thunderbird 2.0.0.6 (Macintosh/20070728) Hi, Thank you for any help or suggestions. Regards, Basavaraj int set_realtime(int period, int computation, int constraint) { struct thread_time_constraint_policy ttcpolicy; struct thread_time_constraint_policy ttcpolicyGet; int ret; ttcpolicy.period=period; // HZ/160 ttcpolicy.computation=computation; // HZ/3300; ttcpolicy.constraint=constraint; // HZ/2200; //ttcpolicy.preemptible=1; ttcpolicy.preemptible=0 ; if ((ret=thread_policy_set(mach_thread_self(), THREAD_TIME_CONSTRAINT_POLICY, (thread_policy_t)&ttcpolicy, THREAD_TIME_CONSTRAINT_POLICY_COUNT)) != KERN_SUCCESS) { fprintf(stderr, "set_realtime() failed with code %d.\n",ret); return 0; } return 1; } and the call is: set_realtime(1250000,60000,90000); Thanks, -n ------------------------------------------------------------------------ _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/basavarajdevelop%40gmail.… _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… I also get this problem. Apple, please suggest some solution to this problem. Why reading data on network sometimes take more time than usually. In my case it is very critical, server drops video and audio packets if client don't receive in allowed time. Problem is more prominent in Linux and Mac OS X, but in Windows, it just works fine. If we launch or do some operation on some cpu intensive application or simple X11 apps, then problem is more frequent. Problem is not because of receiving data. I saw tcp dump on Ethereal, actual problem is with TCP acknowledgment. Acknowledgments are not going immediately after receiving packets. Disabling Naggle algorithm didn't helped. Please help any one or Apple. Nelson Lee wrote: I am currently setting the thread priority of my networking thread as follows and it works! In my thread, I make a networking call using select; more specifically, I am reading from a socket. From timing debug outputs, the reading occurs at regular intervals (~1 ms), but occasionally, the receive call takes 10 ms which throws off my application (it's audio). Is this a networking issue or should setting the priority as real-time take care of this? This email sent to basavarajdevelop(a)gmail.com This email sent to site_archiver(a)lists.apple.com

1 0

Thread Priority?
by Nelson Lee 21 Sep '07

21 Sep '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:reply-to:to:subject:mime-version:content-type; bh=eSxhqCs9R/p51tET6Uo1I2Zi8UdhPowWQ+hNOG3MSto=; b=HaBfeYWUzSWOytVYyHDQz+mF2Tx6xGIZdA9lDwkx8mrpS7yAop+r7r1NMkfR9IwcZaePRHQDltoAZWjg8dZpAWkA4u08wYIXa7EjSjV+MU0zk2hCUwC2W3P2S0FK1tfkP+dqaN7ZFdi0LRo3SjIrLAOETeo6Y08B/51pOF2Bxuo= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:reply-to:to:subject:mime-version:content-type; b=bFfWzJmsPJsn79ywRSBDV3i5RLIyBAfqjHk/pIVnVv/w3igs//nZfSfGoc+II7/+KN6RjbiKxestMiJEbc7DWyGBLfCDv1KSAPrgPY3OIIOIbWlEXTlfv8GzGbXFvn24NNw7dBL9A1ErlBpHqgharcymJxU+bI38rb/dcNGKxrU= _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Posix shared memory - permission problems?
by Tom O'Grady 20 Sep '07

20 Sep '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 20 Sep 2007, at 11:15, Tom O'Grady wrote: Jonas _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/tom.ogrady%40cedaraudio.c… This email sent to tom.ogrady(a)cedaraudio.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Thanks very much Jonas - I think that's (embarrassingly) the solution to the problem! I knew that 0x0777 was the same as 0x777 but I hadn't realised that the chmod bits were octal - when I read back from the fstat structure to 'check' the permissions, my debugger had the st_mode variable read as hexadecimal, so I was under the impression that everything was fine and the mode was indeed set to 0777. Regards, Tom ----- Original Message ----- From: "Jonas Maebe" <jonas.maebe(a)elis.ugent.be> To: <darwin-dev(a)lists.apple.com> Sent: Thursday, September 20, 2007 10:22 AM Subject: Re: Posix shared memory - permission problems? you pass "0x777" for the mode_t value; this lacks the S_IWUSR bit (0000200)--perhaps you intended to use "0777" instead? A failure due to this would match the errno value (EACCES) that you >report for the subsequent O_RDWR attempt... this won't be the problem. that was just mistyping in my e-mail. in my code i definitely use 0x0777 so that user, group and other all have read, write and execute access. 0x0777 is the same as 0x777. The "777" from chmod is not hexadecimal notation (0x) but octal notation (0). So try using "0777" rather than "0x0777" or "0x777". This email sent to site_archiver(a)lists.apple.com

1 0

Re: Posix shared memory - permission problems?
by Jonas Maebe 20 Sep '07

20 Sep '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 20 Sep 2007, at 11:15, Tom O'Grady wrote: Jonas _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… you pass "0x777" for the mode_t value; this lacks the S_IWUSR bit (0000200)--perhaps you intended to use "0777" instead? A failure due to this would match the errno value (EACCES) that you >report for the subsequent O_RDWR attempt... this won't be the problem. that was just mistyping in my e-mail. in my code i definitely use 0x0777 so that user, group and other all have read, write and execute access. 0x0777 is the same as 0x777. The "777" from chmod is not hexadecimal notation (0x) but octal notation (0). So try using "0777" rather than "0x0777" or "0x777". This email sent to site_archiver(a)lists.apple.com

1 0

Re: Posix shared memory - permission problems?
by Tom O'Grady 20 Sep '07

20 Sep '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com hi terry, derek, thanks very much for your replies. to answer both those questions: _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Setting aside your use of the term "shared memory" :-), I notice that in your example: SharedMemID = shm_open(memname, O_CREAT | O_RDWR, 0x777); you pass "0x777" for the mode_t value; this lacks the S_IWUSR bit (0000200)--perhaps you intended to use "0777" instead? A failure due to this would match the errno value (EACCES) that you >report for the subsequent O_RDWR attempt... this won't be the problem. that was just mistyping in my e-mail. in my code i definitely use 0x0777 so that user, group and other all have read, write and execute access. i know that a lot of these permissions are unnecessary but i wanted to allow everything until i got closer to the cause. On Sep 19, 2007, at 8:32 AM, Tom O'Grady wrote: [creating read only shared memory using same name]: CLockMutex mutexSection(mutex); // mutex created using posix semaphore. SharedMemID = shm_open(memname, O_RDONLY, 0x777); // Create shared memory with read access. struct stat fstats; fstat(SharedMemID, &fstats); // check to see if the memory has been allocated and initialised // In this case, the size is already set and the permissions are set to 0x777. Therefore no truncation is required - I just map the memory: smptr = mmap( 0, size, PROT_READ, MAP_SHARED, SharedMemID, 0 ); I suspect that in your real code you are specifying O_CREAT in the O_RDNOLY case, and that is causing the mode to get rest on the shared memory object. If you're not sure if the shared memory exists, you should specify O_EXCL on the create, which will prevent it from being created if it already exists. again - in this test case i am definitely not specifiying O_CREAT when I open the memory for read only access. i only call shm_open with the O_CREAT flag in the first instance, when it's opened and created with read AND write permissions. even though i was under the impression that O_CREAT has no effect when the shared memory exists already, i decided not to call it in the second instance for the simplicity of this test. i'm going to try and strip out the semaphore-implemented mutex to see whether the test works fine then. if it does i'll report back as it might be of use to others. otherwise - i'd still appreciate any suggestions! thanks, tom This email sent to site_archiver(a)lists.apple.com

1 0

Re: Identifying directory changes with kqueue/keven
by Michael Smith 20 Sep '07

20 Sep '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Sep 19, 2007, at 12:04 PM, darwin-dev-request(a)lists.apple.com wrote: I'm trying to watch a directory for changes, so the couple kqueue/ kevent could do the job (aside from the Carbon FNSubscribe). Only problem is that, like FNSubscribe, kqueue/kevent can't tell what file in the directory changed (or maybe I missed something ?). No, that's the normal implementation. The alternatives are all worse. There's also the issue of "what changes actually matter"? = Mike _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… So, is there any simple way to identify the changed file, or do I have to keep track of all files in the directory, and compare this to the directory every time a change occurs ? This could be quite expensive and handling events like file renaming would be a real pain... The system has to be able to make forward progress in the case where a client isn't keeping up with notifications, so for a given watched object it's reasonable to have it report "this changed", but committing to remembering everything that has changed for an arbitrary (and potentially long) period of time is not practical. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Posix shared memory - permission problems?
by Derek Kumar 20 Sep '07

20 Sep '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Posix shared memory - permission problems?
by Terry Lambert 19 Sep '07

19 Sep '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com -- Terry _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… On Sep 19, 2007, at 8:32 AM, Tom O'Grady wrote: [creating read only shared memory using same name]: CLockMutex mutexSection(mutex); // mutex created using posix semaphore. SharedMemID = shm_open(memname, O_RDONLY, 0x777); // Create shared memory with read access. struct stat fstats; fstat(SharedMemID, &fstats); // check to see if the memory has been allocated and initialised // In this case, the size is already set and the permissions are set to 0x777. Therefore no truncation is required - I just map the memory: smptr = mmap( 0, size, PROT_READ, MAP_SHARED, SharedMemID, 0 ); I suspect that in your real code you are specifying O_CREAT in the O_RDNOLY case, and that is causing the mode to get rest on the shared memory object. If you're not sure if the shared memory exists, you should specify O_EXCL on the create, which will prevent it from being created if it already exists. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Identifying directory changes with kqueue/kevent
by Fred Leboucher 19 Sep '07

19 Sep '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Sep 19, 2007, at 10:32 PM, Shaun Wexler wrote: On Sep 19, 2007, at 9:01 AM, Fred Leboucher wrote: Except that my code has to be compatible with 10.3.9 and higher :/ On Sep 19, 2007, at 5:56 PM, Dave Zarzycki wrote: Fred _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… It sounds like you want the new FSEvents API in Leopard. The new FSEvents API is a sustainable interface, forward-compatible from Leopard onwards, however in this case the underlying kernel mechanisms do exist as SPI in earlier OS versions. Simply create a wrapper which calls the legacy SPI's conditionally via runtime introspection. -- Shaun Wexler MacFOH http://www.macfoh.com "Problems cannot be solved by the same level of thinking that created them." - Albert Einstein It exists in Tiger (xnu-792 / http://fxr.watson.org/fxr/source/bsd/ sys/fsevents.h?v=xnu-792), but it looks like it doesn't exist in prior versions. If I'm not mistaken, 10.3.9 has a xnu-517, and fsevents.h is not present in this version (http://fxr.watson.org/fxr/ source/bsd/sys/fsevents.h?v=xnu-517). Are you sure that it's not Tiger-only ? This email sent to site_archiver(a)lists.apple.com

1 0

Re: Identifying directory changes with kqueue/kevent
by Shaun Wexler 19 Sep '07

19 Sep '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Sep 19, 2007, at 9:01 AM, Fred Leboucher wrote: Except that my code has to be compatible with 10.3.9 and higher :/ On Sep 19, 2007, at 5:56 PM, Dave Zarzycki wrote: _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com It sounds like you want the new FSEvents API in Leopard. The new FSEvents API is a sustainable interface, forward-compatible from Leopard onwards, however in this case the underlying kernel mechanisms do exist as SPI in earlier OS versions. Simply create a wrapper which calls the legacy SPI's conditionally via runtime introspection. -- Shaun Wexler MacFOH http://www.macfoh.com "Problems cannot be solved by the same level of thinking that created them." - Albert Einstein smime.p7s

1 0