June 2006 - Darwin-kernel - lists.apple.com

Re: launch agent questions
by Stephane 16 Jun '06

16 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Jun 16, 2006, at 11:49 AM, Finlay Dobbie wrote: I think Dave meant that you should use an alternative to LaunchAgents, such as programmatically modifying the login items, which is supported and should work in future releases of the operating system, current versions of the operating system and previous versions of the operating system. Going forwards, once LaunchAgents work properly (in Leopard), you should think about transitioning to them. At this point in time, it's probably best not to try and implement a LaunchAgent. Let's imagine this technology works fine on Leopard. Let's imagine some software need to be both 10.4 and Leopard compatible. I know backward compatibility is so yesterday these days but... _______________________________________________ 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 16/06/06, Stephane Sudre <ssudre(a)intego.com> wrote: > I'm terribly sorry to report that LaunchAgents are for all practical > purposes, busted in Tiger. I'd advise against using them until > Leopard. Sorry. Stupid question, how could one do that? Since it's a .plist and I don't remember seeing a minimum OS version key, what would be the solution to prevent it from running on 10.4? This is how I read it in the first place. The question was more about backward compatibility: How can you prevent your Launch Agent from being "launched" on 10.4 because of possible bugs? Since it looks like the purpose of launch agent is to just automatically run a task, does that mean that you need to manually run the task at least once to see if you can run it automatically later. This seems to defeat the automatic goal. This email sent to site_archiver(a)lists.apple.com

1 0

Re: may b a minor error
by Terry Lambert 16 Jun '06

16 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Jun 15, 2006, at 4:50 PM, Jeremy Pereira wrote: Correct. I call BS on that. -- 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.… On 15 Jun 2006, at 21:36, Terry Lambert wrote: All of the accessor/mutator functions applicable to uio structures are fairly well documented in <sys/uio.h> itself. A comment in the header like "uio_resid - return the residual IO value for the given uio_t" is *not* "fairly well documented". It's barely adequate given that you could deduce what the function does from its name. The same applies to almost all of the functions declared in that header. I said "fairly well". I also referred you to the OpenDarwin source code. For "badly documented", I'd say it meant "comments do not match code", which is not the case here. For "well documented" (which I didn't claim), you'd need to have attended or otherwise obtained the published materials from the last WWDC, where a number of talks were given and labs were held. There have also been a number of "Developer Kitchens" on various topics on the Apple Campus, which tend to be multiday events concentrating on specific technologies. Great documentation would require that someone write a book (hint to the more literate members of this list who are looking for a project). ...there's also those of us who are on this list, and try to help out by voluntarily answering developer questions, but who have day jobs working on code, and no ability to officially speak for Apple, including but not limited to, commitments to work on documentation instead of code, or ability to make forward looking statements. We also can't disclose Apple confidential WWDC or "developer kitchen" slides, talks, or reference materials. I'm still waiting for the documentation for the VFS KPI more than a year after it was introduced. Frankly, it's just not good enough. Have you filed a bug report/feature request? As a lowly kernel engineer, I'd have to say that an outside bug report is more likely to get a block of time from an overworked technical writer than anything I could personally say to anyone. This email sent to site_archiver(a)lists.apple.com

1 0

Re: may b a minor error
by Michael Smith 16 Jun '06

16 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Jun 15, 2006, at 6:12 AM, Andrew Gallatin wrote: I honestly wasn't trying to bait you, but since you seem to be baiting me... with accessor/mutator functions to insulate the underlying implementation from necessary data structure changes. This is actually a fundamental Computer Science concept for object oriented programming. There's a good Wikipedia article on it at: <http://en.wikipedia.org/wiki/Accessor> Is it found right between "slow" and "inefficent" ? :) Solaris has a much, much more stable driver ABI than you guys do (drivers compiled before MacOSX even existed still work) and it allows direct access to structures (like uio). Sure, and if you want to assume that changes in hardware architecture and the way software uses that hardware stopped fifteen years ago then that model works fine. = Mike _______________________________________________ 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 Jun 15, 2006, at 1:49 PM, Andrew Gallatin wrote: Terry Lambert writes: More than a theory; I've personally made kernel changes in this area for 64 bit support in the Tiger release that would break code that was using promiscuous knowledge of the structure contents, instead of using an accessor/mutator. The idea is to replace data interfaces This would not be required if you had a 64-bit kernel. Of course, if there was a 64-bit kernel, it wouldn't run on 32-bit hardware, and it would require a complete re-write of every device driver in existence. It's fair to say that the folks making these decisions aren't *complete* idiots. Look in the second section, or seqrch in the page for "accessor". This email sent to site_archiver(a)lists.apple.com

1 0

Re: may b a minor error
by Andrew Gallatin 15 Jun '06

15 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Terry Lambert writes: > On Jun 15, 2006, at 6:12 AM, Andrew Gallatin wrote: > > > sugumaran narayanan writes: > >> Hi > >> I am quite new to the Mac world...I am trying to > >> write(using xcode on tiger) a simple character > >> driver(openbsd.org example) and am facing few > >> problems... > >> > >> During compile, struct uio is not recognized and I > >> get a dereferencing pointer to incomplete type > >> error...can somebody help...maybe I am not including > >> the correct libraries... > > > > <....> > > > > MacOSX does not allow access to core kernel structures like > > uio on the theory that Apple may want to change them some day, > > and does not want to break binary drivers when the structure > > changes. > > More than a theory; I've personally made kernel changes in this area I honestly wasn't trying to bait you, but since you seem to be baiting me... > for 64 bit support in the Tiger release that would break code that was > using promiscuous knowledge of the structure contents, instead of > using an accessor/mutator. The idea is to replace data interfaces This would not be required if you had a 64-bit kernel. > with accessor/mutator functions to insulate the underlying > implementation from necessary data structure changes. > > This is actually a fundamental Computer Science concept for object > oriented programming. There's a good Wikipedia article on it at: > > <http://en.wikipedia.org/wiki/Accessor> > > Look in the second section, or seqrch in the page for "accessor". Is it found right between "slow" and "inefficent" ? :) Solaris has a much, much more stable driver ABI than you guys do (drivers compiled before MacOSX even existed still work) and it allows direct access to structures (like uio). Drew _______________________________________________ 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: may b a minor error
by Terry Lambert 15 Jun '06

15 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Jun 15, 2006, at 6:12 AM, Andrew Gallatin wrote: Hi I am quite new to the Mac world...I am trying to write(using xcode on tiger) a simple character driver(openbsd.org example) and am facing few problems... During compile, struct uio is not recognized and I get a dereferencing pointer to incomplete type error...can somebody help...maybe I am not including the correct libraries... <....> MacOSX does not allow access to core kernel structures like uio on the theory that Apple may want to change them some day, and does not want to break binary drivers when the structure changes. <http://en.wikipedia.org/wiki/Accessor> Look in the second section, or seqrch in the page for "accessor". See also: amt = MIN(uio->uio_resid, (echomsg->len - uio->uio_offset > 0) ? For example, in the above code you want to use uio_resid(uio) and uio_offset(uio). Correct. -- 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.… sugumaran narayanan writes: More than a theory; I've personally made kernel changes in this area for 64 bit support in the Tiger release that would break code that was using promiscuous knowledge of the structure contents, instead of using an accessor/mutator. The idea is to replace data interfaces with accessor/mutator functions to insulate the underlying implementation from necessary data structure changes. This is actually a fundamental Computer Science concept for object oriented programming. There's a good Wikipedia article on it at: This is generally called the "KPI". And it was announced as Apple's technical direction in this area at WWDC about 4 years ago, and first widely deployed as part of Tiger. <http://developer.apple.com/documentation/Darwin/Conceptual/KEXTConcept/KEXT… > For a better explanation of the high level rationale behind introducing KPI. All of the accessor/mutator functions applicable to uio structures are fairly well documented in <sys/uio.h> itself. Most other KPI is documented in the header files that define the structures or types that the KPI operates against. If you are in doubt about something, you can look at the internal implementation details by downloading the Darwin sources for the kernel (xnu) from <http://www.opendarwin.org/ >, and looking for the appropriate accessor/mutator functions to use there, or in our published driver sources, or by using the text search field at <http://developer.apple.com>. This email sent to site_archiver(a)lists.apple.com

1 0

Re: may b a minor error
by Andrew Gallatin 15 Jun '06

15 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com sugumaran narayanan writes: > Hi > I am quite new to the Mac world...I am trying to > write(using xcode on tiger) a simple character > driver(openbsd.org example) and am facing few > problems... > > During compile, struct uio is not recognized and I > get a dereferencing pointer to incomplete type > error...can somebody help...maybe I am not including > the correct libraries... <....> MacOSX does not allow access to core kernel structures like uio on the theory that Apple may want to change them some day, and does not want to break binary drivers when the structure changes. This is generally called the "KPI". > amt = MIN(uio->uio_resid, (echomsg->len - > uio->uio_offset > 0) ? For example, in the above code you want to use uio_resid(uio) and uio_offset(uio). Drew _______________________________________________ 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

may b a minor error
by sugumaran narayanan 15 Jun '06

15 Jun '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.com; h=Message-ID:Received:Date:From:Subject:To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=GEaeIAjKfwFXB4VYUl9hfo91CqLWoTCGYQQzG2Efw1KZfjTozIdQdaX+uttVcd1slVrt1C5io8K88Hs2Zox2NphyNSEjmHte6LtB1fq0D5TlAVnzQGFdss1VzqgNc8gRH1b2RHmiip1BD53nZt/E8SmfAjQFETj9bpLn1O/K46o= ; Hi I am quite new to the Mac world...I am trying to write(using xcode on tiger) a simple character driver(openbsd.org example) and am facing few problems... During compile, struct uio is not recognized and I get a dereferencing pointer to incomplete type error...can somebody help...maybe I am not including the correct libraries... my code is as follows #include <mach/mach_types.h> #include <sys/types.h> #include <sys/proc.h> #include <sys/dkstat.h> #include <sys/fcntl.h> //#include <sys/module.h> #include <sys/systm.h> /* uprintf */ #include <sys/errno.h> #include <sys/param.h> /* defines used in kernel.h */ #include <sys/kernel.h> /* types used in module initialization */ #include <sys/conf.h> /* cdevsw struct */ #include <sys/uio.h> /* uio struct */ #include <sys/malloc.h> #include <sys/kernel_types.h> #define MIN(a,b) (((a) < (b)) ? (a) : (b)) #define BUFFERSIZE 256 /* Function prototypes */ d_open_t echo_open; d_close_t echo_close; d_read_t echo_read; d_write_t echo_write; /* Character device entry points */ static struct cdevsw echo_cdevsw = { echo_open, echo_close, echo_read, echo_write, 0, 0, 0, 0, "echo", 33, /* reserved for lkms - /usr/src/sys/conf/majors */ 0, 0, 2, -1 }; struct s_echo { char msg[BUFFERSIZE]; int len; } ; /* vars */ static dev_t sdev; static int count; static struct s_echo *echomsg; //MALLOC_DECLARE(M_ECHOBUF); //MALLOC_DEFINE(M_ECHOBUF, "echobuffer", "buffer for echo module"); kern_return_t chardriver_start (kmod_info_t * ki, void * d) { return KERN_SUCCESS; } kern_return_t chardriver_stop (kmod_info_t * ki, void * d) { return KERN_SUCCESS; } /* * This function is called by the kld[un]load(2) system calls to * determine what actions to take when a module is loaded or unloaded. static int echo_loader(struct module *m, int what, void *arg) { int err = 0; switch (what) { case MOD_LOAD: kldload sdev = make_dev(&echo_cdevsw, 0, UID_ROOT, GID_WHEEL, 0600, "echo"); kmalloc memory for use by this driver MALLOC(echomsg, struct s_echo *, sizeof(struct s_echo), M_ECHOBUF, M_WAITOK); printf("Echo device loaded.\n"); break; case MOD_UNLOAD: destroy_dev(sdev); FREE(echomsg,M_ECHOBUF); printf("Echo device unloaded.\n"); break; default: err = EINVAL; break; } return(err); }*/ int echo_open(dev_t dev, int oflags, int devtype, struct proc *p) { int err = 0; uprintf("Opened device \"echo\" successfully.\n"); return(err); } int echo_close(dev_t dev, int fflag, int devtype, struct proc *p) { uprintf("Closing device \"echo.\"\n"); return(0); } /* * The read function just takes the buf that was saved via * echo_write() and returns it to userland for accessing. * uio(9) */ int echo_read(dev_t dev, struct uio *uio, int ioflag) { int err = 0; int amt=echomsg->len; // struct uio trial; /* * How big is this read operation? Either as big as the user wants, * or as big as the remaining data */ amt = MIN(uio->uio_resid, (echomsg->len - uio->uio_offset > 0) ? echomsg->len - uio->uio_offset : 0); if ((err = uiomove(echomsg->msg + uio->uio_offset,amt,uio)) != 0) { uprintf("uiomove failed!\n"); } return(err); } /* * echo_write takes in a character string and saves it * to buf for later accessing. */ int echo_write(dev_t dev, struct uio *uio, int ioflag) { int err = 0; /* Copy the string in from user memory to kernel memory */ err = copyin(uio->uio_iov->iov_base, echomsg->msg,MIN(uio->uio_iov->iov_len, BUFFERSIZE - 1)); /* Now we need to null terminate, then record the length */ *(echomsg->msg + MIN(uio->uio_iov->iov_len, BUFFERSIZE - 1)) = 0; echomsg->len = MIN(uio->uio_iov->iov_len, BUFFERSIZE); if (err != 0) { uprintf("Write failed: bad address!\n"); } count++; return(err); } regards Sugumaran CL __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.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: mbuf data access (repost)
by Anton Kuzmin 13 Jun '06

13 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com User-agent: Thunderbird 1.5.0.4 (Windows/20060516) Anton To make the copy, you can use mbuf_dup. -josh _______________________________________________ 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.… As I understand, there's one caveat with mbuf_dup: if it's an mbuf on the input path from a network device driver, the data pointer (and length) should be temporarily adjusted to point to the media header, otherwise mbuf_dup doesn't copy it across, which might be useful if there's a need to inject the copy back into the input path. Josh Graessley wrote: Peter is correct. You can use mbuf_mclhasreference to determine if you need to copy the mbuf before modifying the data. This email sent to site_archiver(a)lists.apple.com

1 0

Re: mbuf data access (repost)
by Josh Graessley 13 Jun '06

13 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com To make the copy, you can use mbuf_dup. -josh On Jun 13, 2006, at 11:30 AM, Peter Lovell wrote: On Jun 12, 2006, at 4:46 AM, jy.cases(a)mac.com wrote: On 7 juin 06, at 19:26, Josh Graessley wrote: Is there a way to find out if those mbufs are shared ? Cheers.....Peter _______________________________________________ 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/jgraessley%40apple.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 Peter is correct. You can use mbuf_mclhasreference to determine if you need to copy the mbuf before modifying the data. mbuf_len like mbuf_data works whether or not MBUF_EXT is set. MBUF_EXT just indicates that the actual implementation of the mbuf is storing the data in an external cluster and not in the mbuf data structure itself. The one reason this would be important to a client of the mbuf is the case where you may modify data. In some cases, the reference to the external cluster may be shared. It looks to me as though mbuf_mclhasreference may do what you want. That's in the code but doesn't seem to have a documentation entry. Check bsd/kern/kpi_mbuf.c to see if it's what you need This email sent to jgraessley(a)apple.com smime.p7s

1 0

Re: mbuf data access (repost)
by Peter Lovell 13 Jun '06

13 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Jun 12, 2006, at 4:46 AM, jy.cases(a)mac.com wrote: On 7 juin 06, at 19:26, Josh Graessley wrote: Is there a way to find out if those mbufs are shared ? Cheers.....Peter _______________________________________________ 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.… mbuf_len like mbuf_data works whether or not MBUF_EXT is set. MBUF_EXT just indicates that the actual implementation of the mbuf is storing the data in an external cluster and not in the mbuf data structure itself. The one reason this would be important to a client of the mbuf is the case where you may modify data. In some cases, the reference to the external cluster may be shared. It looks to me as though mbuf_mclhasreference may do what you want. That's in the code but doesn't seem to have a documentation entry. Check bsd/kern/kpi_mbuf.c to see if it's what you need This email sent to site_archiver(a)lists.apple.com

1 0