December 2007 - Darwin-dev - lists.apple.com

Re: DTrace and C++ function
by Dmitry Markman 25 Dec '07

25 Dec '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com thank you very much as I said in my off-list message to you I submitted bugreport to Apple, where I asked question which D pragma will be equivalent for -xmangled dtrace switch. so you answered to my question thank again dm On Dec 24, 2007, at 9:10 PM, Eric Gouriou wrote: Here is an undocumented way. In the DATA section of a custom DTrace instrument, add: #pragma D option mangled Eric Indeed DTrace understands glob patterns, not regular expressions. man dtrace has examples of ObjC probes using such patterns. The idiomatic way to match :: is to use ??. I''ll also add that -xmangled was added to switch DTrace to use the mangled representation instead of the default demangled one. It can be very useful for C++ work. This article might also be of interest: <http://developers.sun.com/solaris/articles/dtrace_cc.html> Although it looks like the default there is to use mangled names. Dmitry Markman _______________________________________________ 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 Dec 22, 2007, at 12:43 PM, Dmitry Markman wrote: do you know how i can set -xmangled from Instruments application? This is the equivalent of -xmangled, from within the script. I am not sure of the consequences of setting this option on other instruments that might be instantiated in the same run. On Dec 22, 2007, at 1:59 PM, Eric Gouriou wrote: On Dec 21, 2007, at 7:02 PM, Dmitry Markman wrote: I should use *test_cpp* Terry mentioned regular expression, so I thought I have to use .*test_cpp.* but that didn'r work This email sent to site_archiver(a)lists.apple.com

1 0

Re: DTrace and C++ function
by Eric Gouriou 25 Dec '07

25 Dec '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Here is an undocumented way. In the DATA section of a custom DTrace instrument, add: #pragma D option mangled Eric Indeed DTrace understands glob patterns, not regular expressions. man dtrace has examples of ObjC probes using such patterns. The idiomatic way to match :: is to use ??. I''ll also add that -xmangled was added to switch DTrace to use the mangled representation instead of the default demangled one. It can be very useful for C++ work. This article might also be of interest: <http://developers.sun.com/solaris/articles/dtrace_cc.html> Although it looks like the default there is to use mangled names. _______________________________________________ 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 Dec 22, 2007, at 12:43 PM, Dmitry Markman wrote: do you know how i can set -xmangled from Instruments application? This is the equivalent of -xmangled, from within the script. I am not sure of the consequences of setting this option on other instruments that might be instantiated in the same run. On Dec 22, 2007, at 1:59 PM, Eric Gouriou wrote: On Dec 21, 2007, at 7:02 PM, Dmitry Markman wrote: I should use *test_cpp* Terry mentioned regular expression, so I thought I have to use .*test_cpp.* but that didn'r work This email sent to site_archiver(a)lists.apple.com

1 0

Re: Identifying CD drives across reboots
by Stephen Hoffman 24 Dec '07

24 Dec '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Organization: HoffmanLabs LLC User-agent: Thunderbird 2.0.0.9 (Macintosh/20071031) _______________________________________________ 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… Stephen F. Booth wrote: I need to identify the drive in order to determine (and store) the drive's read offset. To preempt your next question: I need to do this to accurately access CDDA in a device-independent manner. :) Consider spinning up the media in the drive and ask for the media ISRC, or whatever media-level identification is present on the particular flavor of CD you're working with. Compare this against the expected and known values in your database, and jump to the right sector. Remember that drives won't get insulted if the application chats using a command subset, but the drives do throw a snit if they don't understand what the application has asked, or don't offer the particular command or feature. Worse, some drives can and do become catatonic when they get confused, and some become solidly confused quite easily. I'd not bother looking for a unique drive serial number or trying to aggregate and synthesize such based on model and vendor. What happens if you're dealing with a tower of identical drives, for instance? The ISRC is probably your best bet here, based on your stated goal. Do you have access to the Red, Yellow and Orange books? These and the T10 and T13 documents are soon to be your friends. (And for anyone just getting their tootsies wet in this area, find and start with the INF-8020/SFF-8020 document. 8020 is long retired and long superseded, but it's one of the few good classic introductions to CD operations around.) Drives also tend to be creative about the parts of and the revisions of T10 and T13 implemented. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Identifying CD drives across reboots
by Stephen F. Booth 24 Dec '07

24 Dec '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Being that this is an optical drive, you can't use the expedient of writing something identifiable onto the media. Which begs the question that's been on my mind since this thread started: Since this is removable optical media, why are you trying to identify the drive, rather than the media inserted into the drive? Stephen _______________________________________________ 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 need to identify the drive in order to determine (and store) the drive's read offset. To preempt your next question: I need to do this to accurately access CDDA in a device-independent manner. :) This email sent to site_archiver(a)lists.apple.com

1 0

Re: DTrace and C++ function
by Dmitry Markman 22 Dec '07

22 Dec '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi Eric thanks a lot do you know how i can set -xmangled from Instruments application? thanks On Dec 22, 2007, at 1:59 PM, Eric Gouriou wrote: Indeed DTrace understands glob patterns, not regular expressions. man dtrace has examples of ObjC probes using such patterns. The idiomatic way to match :: is to use ??. I''ll also add that -xmangled was added to switch DTrace to use the mangled representation instead of the default demangled one. It can be very useful for C++ work. This article might also be of interest: <http://developers.sun.com/solaris/articles/dtrace_cc.html> Although it looks like the default there is to use mangled names. Eric Dmitry Markman _______________________________________________ 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 Dec 21, 2007, at 7:02 PM, Dmitry Markman wrote: I should use *test_cpp* Terry mentioned regular expression, so I thought I have to use .*test_cpp.* but that didn'r work This email sent to site_archiver(a)lists.apple.com

1 0

Re: Configure and multiple OS X archs
by Andre-John Mas 22 Dec '07

22 Dec '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 22-Dec-07, at 11:51 , Peter O'Gorman wrote: Hi, #fix_mac_groups #fix_parameters $JPATH/lib/javax.comm.properties if test "x$enable_universal" != xno; then if test -d "/Developer/SDKs/MacOSX10.4u.sdk"; then echo "library os.arch... universal (i386 ppc)" CFLAGS="-O -g -isysroot /Developer/SDKs/MacOSX10.4u.sdk -arch i386 -arch ppc" LDFLAGS="-arch i386 -arch ppc" is much better because, for example, a gentoo user can then do: ./configure CFLAGS="-O3 -funroll-loops" .... or whatever they like :-) Thanks for that. Andre _______________________________________________ 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… Andre-John Mas wrote: I am trying to modify a configure script so that it supports generating of a universal binary (PPC & ix86). What I want to do is ensure that a universal binary is created, when the necessary SDK is available and the user had not explicitly stated they don't want a universal binary. Does this look right: Please do not replace any CFLAGS or LDFLAGS that a user may have specified: CFLAGS="$CFLAGS -arch i386 -arch ppc" LDFLAGS="$LDFLAGS -arch i386 -arch ppc" This email sent to site_archiver(a)lists.apple.com

1 0

Re: DTrace and C++ function
by Eric Gouriou 22 Dec '07

22 Dec '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Indeed DTrace understands glob patterns, not regular expressions. man dtrace has examples of ObjC probes using such patterns. The idiomatic way to match :: is to use ??. I''ll also add that -xmangled was added to switch DTrace to use the mangled representation instead of the default demangled one. It can be very useful for C++ work. This article might also be of interest: <http://developers.sun.com/solaris/articles/dtrace_cc.html> Although it looks like the default there is to use mangled names. Eric _______________________________________________ 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 Dec 21, 2007, at 7:02 PM, Dmitry Markman wrote: I should use *test_cpp* Terry mentioned regular expression, so I thought I have to use .*test_cpp.* but that didn'r work This email sent to site_archiver(a)lists.apple.com

1 0

Re: Configure and multiple OS X archs
by Peter O'Gorman 22 Dec '07

22 Dec '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Thunderbird 2.0.0.9 (X11/20071115) Andre-John Mas wrote: > Hi, > > I am trying to modify a configure script so that it supports generating > of a universal binary (PPC & ix86). What I want to do is ensure that > a universal binary is created, when the necessary SDK is available > and the user had not explicitly stated they don't want a universal > binary. Does this look right: > > #fix_mac_groups > #fix_parameters $JPATH/lib/javax.comm.properties > if test "x$enable_universal" != xno; then > if test -d "/Developer/SDKs/MacOSX10.4u.sdk"; then > echo "library os.arch... universal (i386 ppc)" > CFLAGS="-O -g -isysroot /Developer/SDKs/MacOSX10.4u.sdk > -arch i386 -arch ppc" > LDFLAGS="-arch i386 -arch ppc" Please do not replace any CFLAGS or LDFLAGS that a user may have specified: CFLAGS="$CFLAGS -arch i386 -arch ppc" LDFLAGS="$LDFLAGS -arch i386 -arch ppc" is much better because, for example, a gentoo user can then do: ./configure CFLAGS="-O3 -funroll-loops" .... or whatever they like :-) Peter -- Peter O'Gorman http://pogma.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: DTrace and C++ function
by Dmitry Markman 22 Dec '07

22 Dec '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Oh, well On Dec 21, 2007, at 8:10 PM, Dmitry Markman wrote: Hi Terry thanks a lot for your answer but probably I didn't get it right here is what I did I created simple command line tool #include <iostream> #include <unistd.h> using namespace std; extern "C" { void test_c() { cout << "Hello, World! test_cpp" << endl; } } void test_cpp() { test_c(); } int main (int argc, char * const argv[]) { for(int i = 0; i < 100; i++) { sleep(1); test_cpp(); } return 0; } nm returns ....T __Z8test_cppv and ......T _test_c so I created 2 probs one for test_c pid : DTraceCPPFunction : test_c:entry and another for .*test_cpp.* pid : DTraceCPPFunction : .*test_cpp.* : entry but only prob for test_c did work and I can clearly see in the stack trace that test_cpp was called thanks On Dec 21, 2007, at 5:24 PM, Terry Lambert wrote: -- Terry Hi, is it possible to trace C++ function? I was able to create prob for c++ function ::foo_cpp that was called from c function foo_c I created 2 probs one for foo_c and another for ::foo_cpp but I was able to see events only for foo_c prob (and I can see ::foo_cpp in stack) and none for ::foo_cpp prob I think it's wrong, because foo_c is called ONLY from ::foo_cpp Dmitry Markman _______________________________________________ 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/tlambert%40apple.com Dmitry Markman _______________________________________________ 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/dmarkman%40mac.com Dmitry Markman _______________________________________________ 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 should use *test_cpp* Terry mentioned regular expression, so I thought I have to use .*test_cpp.* but that didn'r work thanks You can't use a colon in a name, but you can use a regular expression. You can also add a string compare as one of the constraints on pthe probe firing (in the /<predicate>/). On Dec 21, 2007, at 7:14 AM, Dmitry Markman wrote: This email sent to tlambert(a)apple.com This email sent to dmarkman(a)mac.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: DTrace and C++ function
by Dmitry Markman 22 Dec '07

22 Dec '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi Terry thanks a lot for your answer but probably I didn't get it right here is what I did I created simple command line tool #include <iostream> #include <unistd.h> using namespace std; extern "C" { void test_c() { cout << "Hello, World! test_cpp" << endl; } } void test_cpp() { test_c(); } int main (int argc, char * const argv[]) { for(int i = 0; i < 100; i++) { sleep(1); test_cpp(); } return 0; } nm returns ....T __Z8test_cppv and ......T _test_c so I created 2 probs one for test_c pid : DTraceCPPFunction : test_c:entry and another for .*test_cpp.* pid : DTraceCPPFunction : .*test_cpp.* : entry but only prob for test_c did work and I can clearly see in the stack trace that test_cpp was called thanks On Dec 21, 2007, at 5:24 PM, Terry Lambert wrote: -- Terry Hi, is it possible to trace C++ function? I was able to create prob for c++ function ::foo_cpp that was called from c function foo_c I created 2 probs one for foo_c and another for ::foo_cpp but I was able to see events only for foo_c prob (and I can see ::foo_cpp in stack) and none for ::foo_cpp prob I think it's wrong, because foo_c is called ONLY from ::foo_cpp Dmitry Markman Dmitry Markman _______________________________________________ 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 can't use a colon in a name, but you can use a regular expression. You can also add a string compare as one of the constraints on pthe probe firing (in the /<predicate>/). On Dec 21, 2007, at 7:14 AM, Dmitry Markman 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/tlambert %40apple.com This email sent to tlambert(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0