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

Re: Total RAM slot count obtained programatically
by Peter Bierman 03 Aug '06

03 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Is there an easy way programmatically to determine how many RAM slots there are and what is filled in them. I ran IORegistryExplorer and it doesn't look like it informs you of the missing slots, just tells you what RAM is installed. I need to know the count of slots their names (DIMM0, DIMM1) and which ones are filled and with what type. IORegistryExplorer almost does it except doesn't inform of the slots with no RAM. It would also be nice if I could find out information as how much each slot supports. man system_profiler system_profiler SPMemoryDataType -pmb _______________________________________________ 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 11:19 AM -0600 8/3/06, matt jaffa wrote: This email sent to site_archiver(a)lists.apple.com

1 0

Total RAM slot count obtained programatically
by matt jaffa 03 Aug '06

03 Aug '06

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:to:subject:mime-version:content-type:content-transfer-encoding:content-disposition; b=CZIbiIvdcr1LcORPK/zhmdrOBilxXfA9dwBHbZyJtCVM64WZjNrnUIzpa0qphnQIds6JSo9Gg6EPK3uygfaTH9TOkc7rZgzdZEUJ+RT1H/d0wlNnj+bNUQ/DHZd1TiY3Q/sqhBzxCAP3F8eZqJSwwikuCXVdfyjjJg/rM7zHns4= Hi, Is there an easy way programmatically to determine how many RAM slots there are and what is filled in them. I ran IORegistryExplorer and it doesn't look like it informs you of the missing slots, just tells you what RAM is installed. I need to know the count of slots their names (DIMM0, DIMM1) and which ones are filled and with what type. IORegistryExplorer almost does it except doesn't inform of the slots with no RAM. It would also be nice if I could find out information as how much each slot supports. Thanks _______________________________________________ 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

Problems with open() and close()
by Cem Karan 03 Aug '06

03 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thanks for any help, Cem Karan boolean_t open_TTY_OSX(char *devicePath, OSX_TTYcontext context) { size_t counter; adjAssert((devicePath != NULL) && (context != NULL)); if ((devicePath == NULL) || (context == NULL)) return false; if (ioctl(context->fileDescriptor, TIOCEXCL) == -1) { ERROR_CLOSE_FD_TTY_OSX_PRIVATE(context->fileDescriptor); return false; } if (fcntl(context->fileDescriptor, F_SETFL, 0) == -1) { ERROR_CLOSE_FD_TTY_OSX_PRIVATE(context->fileDescriptor); return false; } if (fcntl(context->fileDescriptor, F_NOCACHE, 1) == -1) { ERROR_CLOSE_FD_TTY_OSX_PRIVATE(context->fileDescriptor); return false; } for (counter = 0; counter < NCCS; counter++) (context->current).c_cc[counter] = _POSIX_VDISABLE; cfmakeraw(&(context->current)); (context->current).c_cc[VMIN] = 1; (context->current).c_cc[VTIME] = 1; cfsetspeed(&(context->current), B57600); /* Set 57600 baud */ (context->current).c_cflag |= (CS8 | CREAD | CLOCAL); return true; } void close_TTY_OSX(OSX_TTYcontext context) { int result; adjAssert(context != NULL); if (tcdrain(context->fileDescriptor) == -1) { printf("Error waiting for drain - %s(%d).\n", strerror(errno), errno); } if (ioctl(context->fileDescriptor, TIOCNXCL) == -1) { ERROR_CLOSE_FD_TTY_OSX_PRIVATE(context->fileDescriptor); } result = close(context->fileDescriptor); if (result != 0) perror(NULL); } _______________________________________________ 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… Hi all, first time poster to this list, hope that this is right list to post to... I've adapted some code to my own uses (shown below) but have run into an odd problem, and I'm hoping that someone can explain what is going on to me. As a part of my open_TTY_OSX() function, I call ioctl (context->fileDescriptor, TIOCEXCL), which prevents any non-root program from opening the same device. Now this is where it gets odd for me. I am testing all this code using some unit tests that I wrote. The unit tests call open_TTY_OSX() and then balance the call with a call to close_TTY_OSX() on the same file descriptor. Originally, I did NOT have the ioctl(context->fileDescriptor, TIOCNXCL) call in close_TTY_OSX(), and the result was that if I tried to open the same device again, it would fail with EBUSY. So basically: open_TTY_OSX("/dev/ttyp1", foo) // no problem close_TTY_OSX(foo) // no problem open_TTY_OSX("/dev/ttyp1, bar) // Fails with EBUSY Well, I figured that this isn't that big a deal, I just let my process exit nicely, and the restart the debugging session, right? No good. The only way for me to make things play nice is to quit XCode, and even that doesn't seem to work 100% of the time. My question is why? When I close, I reset the attributes, and I close() the descriptor. That should make the blocking part go away, right? What am I doing wrong? Oh, note that adjAssert() is a macro like assert(); I just hacked in some extra stuff that I needed, and didn't what the names conflicting. #define ERROR_CLOSE_FD_TTY_OSX_PRIVATE (FD) \ { \ struct stat myStat; \ perror (NULL); \ fflush (stderr); \ fflush (stdout); \ if (FD != -1) \ { \ printf("Closing file descriptor"); \ fstat(FD, &myStat); \ close (FD); \ } \ } context->fileDescriptor = open(devicePath, O_RDWR | O_NOCTTY | O_NONBLOCK); if (context->fileDescriptor == -1){ ERROR_CLOSE_FD_TTY_OSX_PRIVATE(context->fileDescriptor); return false; } if (tcgetattr(context->fileDescriptor, &(context->original)) == -1) { printf("Error getting tty attributes %s - %s(%d).\n", devicePath, strerror(errno), errno); ERROR_CLOSE_FD_TTY_OSX_PRIVATE(context->fileDescriptor); return false; } memcpy(&(context->current), &(context->original), sizeof(struct termios)); if (tcsetattr(context->fileDescriptor, TCSANOW, &(context- >current)) == -1) { ERROR_CLOSE_FD_TTY_OSX_PRIVATE(context->fileDescriptor); return false; } if (tcsetattr(context->fileDescriptor, TCSANOW, &(context- >original)) == -1) { printf("Error resetting tty attributes - %s(%d).\n", strerror(errno), errno); } This email sent to site_archiver(a)lists.apple.com

1 0

Re: Relative setting of RPATH/RUNPATH to overcome need for LD_LIBRARY_PATH
by almisr 03 Aug '06

03 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com http://plumber.gnu-darwin.org/shared.php?_w2pb=pub/projects/rtool-12 Best Regards, {The Unnamable} _______________________________________________ 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… http://plumber.gnu-darwin.org/shared.php?_w2pb=pub/projects/rtool-12/ samples/relative_path This email sent to site_archiver(a)lists.apple.com

1 0

Should getWorkLoop be thread-safe?
by Howard Gayle 03 Aug '06

03 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Apple's "I/O Kit Fundamentals" guide http://developer.apple.com/documentation/DeviceDrivers/Conceptual/IOKitFund… has an example under Handling Events > Work Loops > Listing 7-1 of overriding getWorkLoop(). The example tries to make getWorkLoop() thread-safe. I have some concerns about the example code, but does getWorkLoop() really need to be thread-safe? I couldn't find any real code that worries about this. The code in AppleRAIDSet.cpp in AppleRAID is pretty typical: IOWorkLoop * AppleRAIDSet::getWorkLoop(void) { // Create a WorkLoop if it has not already been done. if (arSetWorkLoop == 0) { arSetWorkLoop = IOWorkLoop::workLoop(); } return arSetWorkLoop; } Is there a race condition where two different threads could call getWorkLoop(), resulting in the unwanted creation of two different IOWorkLoop objects? _______________________________________________ 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: Relative setting of RPATH/RUNPATH to overcome need for LD_LIBRARY_PATH
by Andreas.Haas＠Sun.COM 03 Aug '06

03 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Thu, 3 Aug 2006, Peter O'Gorman wrote: On Aug 3, 2006, at 7:26 PM, Andreas.Haas(a)Sun.COM wrote: Hi all, I try to figure out the link options for specifying RPATH/RUNPATH under Darwin. I need this to overcome the need of setting LD_LIBRARY_PATH with Grid Engine as described in http://gridengine.sunsource.net/issues/show_bug.cgi?id=2090 Solaris: -R $ORIGIN/../lib Linux: -rpath $ORIGIN/../lib I believe there must be something related for Darwin, but I found no such in ld(1) man page. Ok. Then it seems to be necessary to run install_name_tool -change <libname> @executable_path/<relative_path>/<libname> <executable> Thanks, Peter! Cheers, Andreas _______________________________________________ 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… with other OSes there are link options that allow library pathes be set *relative* to the executable path at the time when the executable is linked through Anyone knowing this? $ORIGIN on solaris and linux allow you to specify the library location relative to the application or library that needs it. For similar options on darwin see the man page for install_name_tool(1) and the dyld release note for information about @executable_path and @loader_path. for each library dependency of an executable where a relative path shall be in effect. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Relative setting of RPATH/RUNPATH to overcome need for LD_LIBRARY_PATH
by Peter O'Gorman 03 Aug '06

03 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Aug 3, 2006, at 7:26 PM, Andreas.Haas(a)Sun.COM wrote: Hi all, I try to figure out the link options for specifying RPATH/RUNPATH under Darwin. I need this to overcome the need of setting LD_LIBRARY_PATH with Grid Engine as described in http://gridengine.sunsource.net/issues/show_bug.cgi?id=2090 Solaris: -R $ORIGIN/../lib Linux: -rpath $ORIGIN/../lib I believe there must be something related for Darwin, but I found no such in ld(1) man page. Peter _______________________________________________ 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… with other OSes there are link options that allow library pathes be set *relative* to the executable path at the time when the executable is linked through Anyone knowing this? $ORIGIN on solaris and linux allow you to specify the library location relative to the application or library that needs it. For similar options on darwin see the man page for install_name_tool(1) and the dyld release note for information about @executable_path and @loader_path. This email sent to site_archiver(a)lists.apple.com

1 0

Relative setting of RPATH/RUNPATH to overcome need for LD_LIBRARY_PATH
by Andreas.Haas＠Sun.COM 03 Aug '06

03 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi all, I try to figure out the link options for specifying RPATH/RUNPATH under Darwin. I need this to overcome the need of setting LD_LIBRARY_PATH with Grid Engine as described in http://gridengine.sunsource.net/issues/show_bug.cgi?id=2090 Solaris: -R $ORIGIN/../lib Linux: -rpath $ORIGIN/../lib I believe there must be something related for Darwin, but I found no such in ld(1) man page. Anyone knowing this? Regards, Andreas _______________________________________________ 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… with other OSes there are link options that allow library pathes be set *relative* to the executable path at the time when the executable is linked through This email sent to site_archiver(a)lists.apple.com

1 0

Re: Get the caller of socket_lock()
by Laurent Dumont 02 Aug '06

02 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Microsoft-Entourage/11.1.0.040913 I think this is because the source you have for the socket structure is for the ppc version, and gdb uses that to find the offsets and fields for everything. You could try replacing the reserved3/reserved4 by the new fields in the copy of the sourcefile xnu/bsd/sys/socketvar.h that is used by gdb to determine the structure format. If that fails, you may have to look at the structure in hex with gdb and figure out the last part by hand. Those fields are used for internal debug and not part of the KPI, which provides accessors to certain fields of the socket structure. That's why the format is not in sync with the sources you have... on 8/2/06 13:49, Michael Tuexen at Michael.Tuexen(a)lurchi.franken.de wrote: > Hi Laurent, > > hmmm, sounds really good, but we do not see the symbols you mention > being part of the socket structure... How can we access the fields > in gdb? > > Thank you very much in advance. > > Best regards > Michael > > On Aug 2, 2006, at 10:23 PM, Laurent Dumont wrote: > >> Michael, >> >> On 10.4.x for i386 there is a slightly different debug mechanism >> for the >> socket lock and unlock. Instead of reserved3 and reserverd4, there is >> circular buffers with the 4 last lock/unlocks for the socket: >> >> u_int32_t lock_lr[4]; >> int next_lock_lr; >> u_int32_t unlock_lr[4]; >> int next_unlock_lr; >> >> "next_lock_lr" and "next_unlock_lr" provide the number (0 to 3) for >> the next >> debug link register slot to be used. You can work your way from n-1 >> and find >> the last four lock/unlock recorded for that socket. >> >> Laurent >> >> >> >> on 8/2/06 11:23, Michael Tuexen at Michael.Tuexen(a)lurchi.franken.de >> wrote: >> >>> Dear all, >>> >>> we are currently debugging a locking issue on Mac OS X on Intel. It >>> is related to >>> socket locks and we are able to break into the debugger to inspect >>> the socket >>> structures. On PPC one can find the last caller of the socket_lock() >>> function, >>> but is there a similar way in doing it in Intel, possibly by >>> inspecting the reserved3 >>> or reserved4 fields in the socket structure? What is stored in these >>> fields? >>> >>> Thank you very much for your help in advance. >>> >>> Best regards >>> Michael >>> >>> _______________________________________________ >>> 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/laurent%40apple.com >>> >>> This email sent to laurent(a)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: Get the caller of socket_lock()
by Michael Tuexen 02 Aug '06

02 Aug '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi Laurent, hmmm, sounds really good, but we do not see the symbols you mention being part of the socket structure... How can we access the fields in gdb? Thank you very much in advance. Best regards Michael On Aug 2, 2006, at 10:23 PM, Laurent Dumont wrote: Michael, u_int32_t lock_lr[4]; int next_lock_lr; u_int32_t unlock_lr[4]; int next_unlock_lr; Laurent Dear all, we are currently debugging a locking issue on Mac OS X on Intel. It is related to socket locks and we are able to break into the debugger to inspect the socket structures. On PPC one can find the last caller of the socket_lock() function, but is there a similar way in doing it in Intel, possibly by inspecting the reserved3 or reserved4 fields in the socket structure? What is stored in these fields? Thank you very much for your help in advance. Best regards Michael _______________________________________________ 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/laurent%40apple.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… On 10.4.x for i386 there is a slightly different debug mechanism for the socket lock and unlock. Instead of reserved3 and reserverd4, there is circular buffers with the 4 last lock/unlocks for the socket: "next_lock_lr" and "next_unlock_lr" provide the number (0 to 3) for the next debug link register slot to be used. You can work your way from n-1 and find the last four lock/unlock recorded for that socket. on 8/2/06 11:23, Michael Tuexen at Michael.Tuexen(a)lurchi.franken.de wrote: This email sent to laurent(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0