November 2004 - Darwin-dev - lists.apple.com

Re: Asm question (port from linux ppc)
by Jonas Maebe 27 Nov '04

27 Nov '04

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com On Nov 27, 2004, at 05:03, darwin(a)qbjnet.com wrote: $cat setsp.s .globl _setstack _setstack: lis 1, newstackptr@ha lwz 1, newstackptr@l(1) blr .align 4 % cc setsp.S setsp.s:5:Parameter syntax error (parameter 1) setsp.s:5:Invalid mnemonic 'ha' setsp.s:6:Parameter syntax error (parameter 2) setsp.s:6:Invalid mnemonic 'l(r1)' --- .text .globl _setstack _setstack: lis r3,ha16(_newstackptr) lwz r1,lo16(_newstackptr)(r3) blr --- --- .text .globl _setstack _setstack: lis r3,ha16(L_newstackptr$non_lazy_ptr) lwz r3,lo16(L_newstackptr$non_lazy_ptr)(r3) lwz r1,0(r3) blr .data .non_lazy_symbol_pointer L_newstackptr$non_lazy_ptr: .indirect_symbol _newstackptr .long 0 --- 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… What the correct code on Darwin is depends. If newstackptr is simply a global variable in another object file and you're compiling without PIC, then you can use code like this: (note that r1 must always be valid and thus mist always changed using a single instruction under any PPC ABI I know of; your original linux code is wrong in that respect) If it's a variable imported from another shared library or if you're compiling PIC code, then you need something like this: In general, check the output of "gcc -S somefile.c" to see how things are done under Darwin/Mac OS X. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Asm question (port from linux ppc)
by Justin Walker 27 Nov '04

27 Nov '04

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com On Nov 26, 2004, at 20:03, darwin(a)qbjnet.com wrote: Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | When LuteFisk is outlawed | Only outlaws will have | LuteFisk *--------------------------------------*-------------------------------* _______________________________________________ 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 trying to port an app from linux PPC to MacOSX. There is a short (3 lines) routine in assembly that changes the stack pointer. The code assembles fine in linux with both gcc 2.95 and 3.3.4 but Darwin gcc 3.1 doesn't like it. I tried compiling a couple of short test programs to assembler on both platforms and looked at the generated code. It looked pretty different. Any ideas?? The code just moves a new stack pointer from a global variable into the stack pointer register (1) and returns. I think the assembler for darwin has a different syntax than the Linux version. Check, e.g., the ".s" files in the kernel source (xnu) at the darwin website (<http://developer.apple.com/darwin>). This email sent to site_archiver(a)lists.apple.com

1 0

Asm question (port from linux ppc)
by darwin＠qbjnet.com 27 Nov '04

27 Nov '04

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com I trying to port an app from linux PPC to MacOSX. There is a short (3 lines) routine in assembly that changes the stack pointer. The code assembles fine in linux with both gcc 2.95 and 3.3.4 but Darwin gcc 3.1 doesn't like it. I tried compiling a couple of short test programs to assembler on both platforms and looked at the generated code. It looked pretty different. Any ideas?? The code just moves a new stack pointer from a global variable into the stack pointer register (1) and returns. $cat setsp.s .globl _setstack _setstack: lis 1, newstackptr@ha lwz 1, newstackptr@l(1) blr .align 4 % cc setsp.S setsp.s:5:Parameter syntax error (parameter 1) setsp.s:5:Invalid mnemonic 'ha' setsp.s:6:Parameter syntax error (parameter 2) setsp.s:6:Invalid mnemonic 'l(r1)' -- bob _______________________________________________ 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: IOKit + display models
by Joe van Tunen 27 Nov '04

27 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AcTT6i+uBx/HMRUSTCi4zYnZYTaGQgAEFJbk Thread-topic: IOKit + display models > ---------- > From: Alex Reynolds > Reply To: bio-computing(a)sas.upenn.edu > Sent: Thursday, November 25, 2004 7:54 PM > To: darwin-dev(a)lists.apple.com > Subject: IOKit + display models > > I'd like to figure out how to get from the IOKit call-equivalent of: > > [Sond] reynolda$ ioreg -c AppleDisplay | grep DisplayVendorID > | | | "DisplayVendorID" = 1552 > | | | "DisplayVendorID" = 4268 > [Sond] reynolda$ > > to the corresponding list of the two monitor models connected to my machine, > which the Display system preference says are an "Apple Studio Display" and a > "Dell 1504FP". > > The files in /System/Library/Displays/Overrides/DisplayVendorID-610 are not in > a > consistent format as far as the information they return. I'll bet that isn't > where this information is kept anyway. > > There's an entry in there that matches 1552 to an Apple Studio Display but I > can't find anything for non-Apple displays (e.g. matching the 4268 entry to the > Dell display). > > (This table must be *somewhere* if the Display system preference knows it.) > > Does anyone know how to bend IOKit's ear to return this information? > > TIA, > Alex > I think IODisplayCreateInfoDictionary does everything you need. The source code is available in the IOKitUser Darwin project so you can see how it works. IODisplayTest.c in the same project (but possibly only in Darwin 6.0 or earlier) shows how to use it. Below is an updated version of the IODisplayTest code. I couldn't figure out how to get CFBundleCopyPreferredLocalizationsFromArray or CFBundleCopyLocalizationsForPreferences to work properly so I hard coded it to get the english name. /* cc -o /tmp/iodt graphics.subproj/IODisplayTest.c -framework CoreGraphics -Wall -undefined warning -g */ #include <CoreFoundation/CoreFoundation.h> #include <CoreGraphics/CoreGraphics.h> #include <IOKit/graphics/IOGraphicsLib.h> #include <stdlib.h> void KeyArrayCallback(const void *key, const void *value, void *context) { CFArrayAppendValue(context, key); } int main(int argc, char * argv[]) { io_service_t service; CFDictionaryRef dict; CFDictionaryRef names; CFMutableArrayRef langKeys; CFMutableArrayRef orderLangKeys; CFNumberRef num; CGError err; SInt32 numValue; int i; CGDisplayCount max; CGDirectDisplayID displayIDs[8]; err = CGGetOnlineDisplayList(8, displayIDs, &max); if(err != kCGErrorSuccess) exit(1); if(max > 8) max = 8; for(i = 0; i < max; i++ ) { service = CGDisplayIOServicePort( displayIDs[i] ); if ( !service ) continue; dict = IODisplayCreateInfoDictionary(service, kNilOptions ); if( dict) { printf("Display %d\n", i); num = CFDictionaryGetValue(dict, CFSTR(kDisplayVendorID)); if(num) { CFNumberGetValue(num, kCFNumberSInt32Type, &numValue); printf("vendorID %ld\n", numValue); } num = CFDictionaryGetValue(dict, CFSTR(kDisplayProductID)); if(num) { CFNumberGetValue(num, kCFNumberSInt32Type, &numValue); printf("productID %ld\n", numValue); } names = CFDictionaryGetValue(dict, CFSTR(kDisplayProductName)); langKeys = CFArrayCreateMutable(kCFAllocatorDefault, 0, &kCFTypeArrayCallBacks); CFDictionaryApplyFunction(names, KeyArrayCallback, (void *) langKeys); #if 0 /* doesn't work for command line tools? (always returns only the last langKey: zh_CN = Traditional Chinese) */ orderLangKeys = CFBundleCopyPreferredLocalizationsFromArray(langKeys); #elif 0 /* doesn't work for command line tools? (always returns only the last langKey: zh_CN = Traditional Chinese) */ orderLangKeys = CFBundleCopyLocalizationsForPreferences(langKeys, NULL); #else orderLangKeys = CFArrayCreateMutable(kCFAllocatorDefault, 0, &kCFTypeArrayCallBacks); CFArrayAppendValue(orderLangKeys, CFSTR("en_US")); #endif CFRelease(langKeys); if(orderLangKeys && CFArrayGetCount(orderLangKeys)) { char cName[256]; CFStringRef langKey; CFStringRef localName; langKey = CFArrayGetValueAtIndex(orderLangKeys, 0); localName = CFDictionaryGetValue(names, langKey); CFStringGetCString(localName, cName, sizeof(cName), CFStringGetSystemEncoding()); printf("local name \"%s\"\n", cName); } CFRelease(orderLangKeys); printf("\nAll info:\n"); CFShow(dict); CFRelease(dict); } // -- /* { IOCFPlugInInterface ** interface; IODisplayCreateInterface( service, kNilOptions, kIODisplayControlInterfaceID, &interface ); } */ /* { CFMutableDictionaryRef dict; CFNumberRef num; SInt32 value; dict = CFDictionaryCreateMutable( kCFAllocatorDefault, 0, &kCFTypeDictionaryKeyCallBacks, &kCFTypeDictionaryValueCallBacks ); for( value = 0; value < 0x60; value++) { num = CFNumberCreate( kCFAllocatorDefault, kCFNumberSInt32Type, &value ); CFDictionarySetValue( dict, CFSTR("contrast"), num ); CFRelease(num); err = IODisplaySetParameters( service, kNilOptions, dict ); } CFRelease(dict); } */ // -- } exit(0); return(0); } _______________________________________________ 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: IOKit + display models
by Dan Bernstein 26 Nov '04

26 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:references; b=Mns/0XPTywaOO0EJ5TLQkb7Fr/r387el1M34Ojq53K72fsgM1wtOhX0pPVLQKdsrn27uNEdgRKzFh0noxoflCjE0hJFXcVkekbw/oF0/pJ10/+rftO8XaQCxr/kz6k9uhcd+kPVw3OrfwCk1ti0Sc48PyeWeYbGhQU/EeiAOS9w= The product name could be coming from the EDID block (it is exported in the properties dictionary of the IODisplay (or subclass) object of your display, so you can see it in IORegistryExplorer or ioreg; it is at offset... hmm... 77?). On Thu, 25 Nov 2004 22:54:59 -0500, Alex Reynolds <reynolda(a)sas.upenn.edu> wrote: > > > I'd like to figure out how to get from the IOKit call-equivalent of: > > [Sond] reynolda$ ioreg -c AppleDisplay | grep DisplayVendorID > | | | "DisplayVendorID" = 1552 > | | | "DisplayVendorID" = 4268 > [Sond] reynolda$ > > to the corresponding list of the two monitor models connected to my machine, > which the Display system preference says are an "Apple Studio Display" and a > "Dell 1504FP". > > The files in /System/Library/Displays/Overrides/DisplayVendorID-610 are not in > a > consistent format as far as the information they return. I'll bet that isn't > where this information is kept anyway. > > There's an entry in there that matches 1552 to an Apple Studio Display but I > can't find anything for non-Apple displays (e.g. matching the 4268 entry to the > Dell display). > > (This table must be *somewhere* if the Display system preference knows it.) > > Does anyone know how to bend IOKit's ear to return this information? > > TIA, > Alex > _______________________________________________ > 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/dan.bernstein%40gmail.com > > > > This email sent to dan.bernstein(a)gmail.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

IOKit + display models
by Alex Reynolds 26 Nov '04

26 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Internet Messaging Program (IMP) 3.2.2 I'd like to figure out how to get from the IOKit call-equivalent of: [Sond] reynolda$ ioreg -c AppleDisplay | grep DisplayVendorID | | | "DisplayVendorID" = 1552 | | | "DisplayVendorID" = 4268 [Sond] reynolda$ to the corresponding list of the two monitor models connected to my machine, which the Display system preference says are an "Apple Studio Display" and a "Dell 1504FP". The files in /System/Library/Displays/Overrides/DisplayVendorID-610 are not in a consistent format as far as the information they return. I'll bet that isn't where this information is kept anyway. There's an entry in there that matches 1552 to an Apple Studio Display but I can't find anything for non-Apple displays (e.g. matching the 4268 entry to the Dell display). (This table must be *somewhere* if the Display system preference knows it.) Does anyone know how to bend IOKit's ear to return this information? TIA, Alex _______________________________________________ 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: Simple BSD Sockets question
by Pelle Johansson 25 Nov '04

25 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com 2004-11-25 kl. 00.20 skrev Paul Forgey: _______________________________________________ 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… In my experience, it's a lot more reliable to close the socket unless the signal handler exits the process. Taking a sideways exit out of the process is fine for simple things, but gets ugly if you need to cleanly shut things down (like database files), especially if you have threads. If you depend on accept failing with EINTR, the signal may not arrive while blocking on the accept call. If you happen to get the signal just before or after the accept call, you won't notice. Polling a global shutdown flag set in the signal handler still doesn't help because you may get the signal after checking the flag but before blocking in the accept call. In your case, you could just close the socket in the signal handler. I might have misunderstood the OP, but how would you close a socket if you are currently in a blocking accept call? And if done from another thread, the problem is once you have closed an fd, it becomes available for open again, and you have a potential race issue. I wouldn't recommend closing a fd from another thread than which is using it, but if you feel you have to, make sure you invalidate the variable the other thread is using before the close. I'm unsure if this was a multithreaded application though, and if not I'd go with either closing the socket and invalidating the fd from the signal handler. Even better would probably be to not even enter the accept call the last time, but perhaps that's not possible. Perhaps the OP can tell us how the process knows when it's supposed to quit? Once you get past that first simple socket usage, you should probably be using select(2) or kqueue(2) to wait for data. That way you will only call the system calls when there's data available in them, and they won't block. You probably shouldn't spawn one thread per connection unless you have special needs. -- Pelle Johansson <morth(a)morth.org> This email sent to site_archiver(a)lists.apple.com

1 0

Re: SDKs and linking
by Sam Vaughan 24 Nov '04

24 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thanks for this Peter, NEXT_ROOT does exactly what we want. Sam On 25/11/2004, at 7:09 AM, Peter O'Gorman 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… DYLD_LIBRARY_PATH has nothing whatsoever to do with the linker. I would have thought that setting the NEXT_ROOT environment variable (which xcode does as far as I know) would fix this. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Simple BSD Sockets question
by Paul Forgey 24 Nov '04

24 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In your case, you could just close the socket in the signal handler. On Nov 24, 2004, at 9:56 AM, Pelle Johansson 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/paulf%40aphrodite.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… In my experience, it's a lot more reliable to close the socket unless the signal handler exits the process. Taking a sideways exit out of the process is fine for simple things, but gets ugly if you need to cleanly shut things down (like database files), especially if you have threads. If you depend on accept failing with EINTR, the signal may not arrive while blocking on the accept call. If you happen to get the signal just before or after the accept call, you won't notice. Polling a global shutdown flag set in the signal handler still doesn't help because you may get the signal after checking the flag but before blocking in the accept call. Send the process a signal you are catching, the system call will then interrupt and return. -- Pelle Johansson <morth(a)morth.org> This email sent to paulf(a)aphrodite.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: Simple BSD Sockets question
by Pelle Johansson 24 Nov '04

24 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com 2004-11-24 kl. 16.31 skrev Mark Gilbert: Folks I am communicating between CPUs using a socket. On the server side, I bind, then listen, then accept. On the client, I talk to the server through the socket This is all working perfectly, but I have one small problem How do I release the accept so I can terminate my thread and quit ? _______________________________________________ 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… - When I quit, I need to unblock the last call to accept. (I am using a blocking call). I am sure this is really simple, but I am new to sockets, please give me any pointers (or handles...:-) Send the process a signal you are catching, the system call will then interrupt and return. -- Pelle Johansson <morth(a)morth.org> This email sent to site_archiver(a)lists.apple.com

1 0