August 2006 - Darwin-dev - lists.apple.com

Re: Mount Points
by Dan Shoop 26 Aug '06

26 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Folks. So, I have 2 questions: A directory. The directory used as the mountpoint in the fileysystem. Remove rogoue mountpoints. -- -dhan ------------------------------------------------------------------------ Dan Shoop AIM: iWiring Systems & Networks Architect http://www.ustsvs.com/ shoop(a)iwiring.net http://www.iwiring.net/ 1-714-363-1174 "The wise man doesn't give the right answers, he poses the right questions." -- Claude Levi-Strauss ------------------------------------------------------------------------ iWiring provides systems and networks support for Mac OS X, unix, and Open Source application technologies at affordable rates. _______________________________________________ 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… At 6:02 PM +0100 8/26/06, Mark Gilbert wrote: I am seeing a strange behaviour sometimes when mounting afp network volumes on an Intel iMac. Essentially the OS is creating more than one /Volumes representation for the drive, depending on what steps have occurred, and its creating a persistent entry in /Volumes which I can't 'unmount' Yes, it's probably a real directory. So it can't be unmounted. These are commonly referred to ass mountpoints. They are normally automagically created and destroyed when the OS mounts a volume. Sometimes you'll lose a mount and a mountpoint won't be destroyed. You'll have to rm it manually. 1) What is this phantom directory ? 2) How do I solve my problem and avoid this in the field ? This email sent to site_archiver(a)lists.apple.com

1 0

Mount Points
by Mark Gilbert 26 Aug '06

26 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Folks. So, I have 2 questions: 1) What is this phantom directory ? 2) How do I solve my problem and avoid this in the field ? Cheers Mark. -- mark(a)gallery.co.uk Tel: +44 208 340 5677 fax: +44 870 055 7790 http://www.gallery.co.uk _______________________________________________ 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 am seeing a strange behaviour sometimes when mounting afp network volumes on an Intel iMac. Essentially the OS is creating more than one /Volumes representation for the drive, depending on what steps have occurred, and its creating a persistent entry in /Volumes which I can't 'unmount' For example: I have another Mac sharing its hard disc. This volume is called "G4_HD". When I try to mount it on the iMac, it ends up being referred to as G4_HD-1 by Unix (in /Volumes). This is because there is already a 'persistent' entry in /Volumes called 'G4_HD' from some earlier life. This persistent entry only contains one of the directories which belong to the volume. The problem I have is that I am using file:///Volumes/Drive/Folder/File style URLs to pass pointers to files over a network connection. When I use CFURL to resolve a URL like: file:///Volumes/G4_HD/folder/file it fails, because although the 'real' G4_HD *IS* mounted on the Mac desktop, its actually referred to deep down as G4_HD-1, and 'folder' doesn't exist in the 'phantom' tree that is actually taking up the volume name G4_HD. If I unmount G4_HD in the finder, G4_HD-1 goes from /Volumes, but G4_HD remains...!!!!! I can see another 'phantom' directory from another computer on the network which must have once been mounted here (but isnt anymore) This email sent to site_archiver(a)lists.apple.com

1 0

Re: MACFORGE DOWN
by Kevin Van Vechten 24 Aug '06

24 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com - Kevin On Aug 24, 2006, at 8:31 AM, almisr wrote: :D _______________________________________________ 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 for reporting this -- we're aware of the issue and will be installing more RAM in the web-server today. Fatal error: out of dynamic memory in yy_create_buffer() in Unknown on line 0 it seems that you are trying to create something much larger than was originally anticipated and you have exhausted memory Best Regards {|} This email sent to site_archiver(a)lists.apple.com

1 0

Re: Passing too many parameters to a function?
by seebs＠plethora.net 24 Aug '06

24 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <73491B49-ADEA-4F6E-A7D5-7B0AD31582E2(a)gmx.net>, Uli Kusterer writes: > Does that mean it's actually safe to do this after all, under these >limited circumstances? I originally thought so, as I actually once >managed to call a function with too many parameters (it was a >horrible mess of typecasts -- don't ask, I was young and needed the >experience) and it didn't blow up in my face right away, which kinda >surprised me. But I just thought I'd been lucky. I have no idea. The problem is that there are so very many ways in which it MIGHT fail, depending on exactly what you're passing, and I don't know the ABI. I would consider it almost certainly foolhardy, but it's up to you. >And we want new & old plugins to work on the host app now, while the >future host app will only accept new ones. It sounds like this will >work after all? Don't worry about portability, because even if a port >was planned, it'd require the new callback style *with* the context >pointer right away. In that case, my suggestion is this: Test it. You don't even care about the general case of extra parameters; you care about the SPECIFIC case of this particular extra parameter, and of course, you can check for that. I would suggest, if you wanna be very careful: Embed some assembly around the call to check all the registers before and after to see what happens, just in case of madness. -s _______________________________________________ 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

MACFORGE DOWN
by almisr 24 Aug '06

24 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com :D Best Regards {|} _______________________________________________ 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… Fatal error: out of dynamic memory in yy_create_buffer() in Unknown on line 0 it seems that you are trying to create something much larger than was originally anticipated and you have exhausted memory This email sent to site_archiver(a)lists.apple.com

1 0

Re: Passing too many parameters to a function?
by Uli Kusterer 24 Aug '06

24 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Can you show an example, where sending a superflous parameter to a C function creates a problem with current Mach-O ABIs ? I.e.: Old plugin: callback(); New plugin: callback( contextPtr ); Host app now: callback() { ... } Host app future: callback( Context* contextPtr ) { ... } Cheers, -- M. Uli Kusterer http://www.zathras.de Am 24.08.2006 um 00:34 schrieb Peter Seebach: In message <87994A64-D80B-4C61-85B9-9D50D980B60E@mulle- kybernetik.com>, Nat! wr ites: No; I misinterpreted the question as being about safety and portability in general, rather than about specific API portability. Does that mean it's actually safe to do this after all, under these limited circumstances? I originally thought so, as I actually once managed to call a function with too many parameters (it was a horrible mess of typecasts -- don't ask, I was young and needed the experience) and it didn't blow up in my face right away, which kinda surprised me. But I just thought I'd been lucky. I'm fully aware that there may be some other OSs and platforms and ABIs where this wouldn't be valid, and it's definitely not ANSI C. I'd merely need this for backwards compatibility with plugins and a host app that will run on PowerPC and Intel so they're prepared for handing us the additional parameter we need right now when we're introducing the new Mach-O plugin scheme, even though the host app isn't yet using the additional parameter (and can't require it because legacy plugins don't yet know about the parameter. And we want new & old plugins to work on the host app now, while the future host app will only accept new ones. It sounds like this will work after all? Don't worry about portability, because even if a port was planned, it'd require the new callback style *with* the context pointer right away. PS - Plum/Almisr/whatever your name is: Thanks for the ASCII ribbon. It reminded me I'd have to check my reinstalled mail app whether it generated HTML e-mail. It did :-( Sorry. _______________________________________________ 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: Passing too many parameters to a function?
by almisr 24 Aug '06

24 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Last I knew, that was a part of the C language standard. yep except if u build an known array (each time) -> arg[0] = [type:value] -> arg[1] = [type:value] func (int num,arg[]) parsing args for num yep standard case /* void simple_printf( char *fmt,mixed array); */ void simple_printf( char *fmt,...); Best Regards, {Plum} _ /o\ // \\ The ASCII \\ // Ribbon Campaign \V/ Against HTML /A\ eMail! // \\ _______________________________________________ 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 is not. There is no guarantee that you can pass extra arguments safely. It is not. There is no guarantee that you can pass extra arguments safely. ------------------------------------------------------------------------ -------------------------------------- #include <stdio.h> #include <stdarg.h> int main(int argc, const char * argv[]) { simple_printf("\n%s %s%s\n\n","hello","world","!"); simple_printf("\n%s %s%s %s %s %s %i\n \n","hello","world","!","we","are","in",2006); return 0; } void simple_printf(char *format,...) { va_list list; char *explode; char *print_char; int print_int; va_start(list,format); /*explode format*/ for(explode = format;*explode;++explode) { /* searching % */ if(*explode != '%') putc(*explode,stdout); else { /* ++next */ switch(*++explode) { case 's': { print_char = va_arg(list,char *); printf("%s",print_char); continue; } case 'i': { print_int = va_arg(list,int); printf("%i",print_int); continue; } /* add types */ default:putc(*explode,stdout); } } va_end(list); fflush(stdout); } } ------------------------------------------------------------------------ -------------------------------------- This email sent to site_archiver(a)lists.apple.com

1 0

Re: Passing too many parameters to a function?
by seebs＠plethora.net 23 Aug '06

23 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <36855455-21CD-46C2-9D0E-D9F1D263092C(a)apple.com>, Dave Zarzycki writ es: >Last I knew, that was a part of the C language standard. It is not. There is no guarantee that you can pass extra arguments safely. There is a special guarantee about the calling arguments of main; the startup code has to magically Do The Right Thing. >And it works too. Lots of Unix APIs depends on that. Off the top of my >head, the third argument of open() is optional and signal handers were >originally defined to take one argument, but now they take three. Old >handlers will obviously work, since they ignore the second and third >arguments. Both of those are POSIX, rather than C. The third argument to open isn't optional; it's new, but calls without it are treated just like calls with it, where the third argument is, well, whatever happens to be handy. Maybe a bit of stack, maybe a scratch register. No guarantees! With signals, the arguments are passed, but old code that doesn't check for them will probably compile correctly; there's no guarantee, and you can't make a program that does this without invoking undefined behavior somewhere. -s _______________________________________________ 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: Passing too many parameters to a function?
by seebs＠plethora.net 23 Aug '06

23 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <87994A64-D80B-4C61-85B9-9D50D980B60E(a)mulle-kybernetik.com>, Nat! wr ites: >Can you show an example, where sending a superflous parameter to a C >function creates a problem with current Mach-O ABIs ? No; I misinterpreted the question as being about safety and portability in general, rather than about specific API portability. -s _______________________________________________ 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: Passing too many parameters to a function?
by Ilfak Guilfanov 23 Aug '06

23 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Organization: Datarescue sa/nv Hello Uli, Sunday, August 20, 2006, 12:17:39 PM, you wrote: UK> you know how main() can be declared with varying parameter counts, UK> like UK> int main() UK> or UK> int main( int argc, const char* argv* ) UK> and even more. Does this work for any function? If I pass a function UK> more parameters than it actually uses, is that safe with OS X's UK> current Mach-O ABIs? T think there is no danger in passing more parameters than the called function expects (unless you specify -mrtd flag to the compiler or do something else which changes the default environment). -- Best regards, Ilfak mailto:ig@datarescue.be _______________________________________________ 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