August 2005 - Darwin-dev - lists.apple.com

Re: getpwuid memory leak?!?
by Martin Bestmann 03 Aug '05

03 Aug '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 03.08.2005, at 23:40, Marc Majka wrote: -- Marc Majka On 3 Aug, 2005, at 10:16, Martin Bestmann wrote: Hi, getpwuid Here is a stack crawl returned by "ObjectAlloc" getpwuid getpw getpw_internal copy_user strdup malloc malloc_zone_malloc getpwuid getpw getpw_internal copy_user calloc I'm running on Mac OS X 10.4.2 Thanks, Martin _______________________________________________ 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/majka%40apple.com This email sent to majka(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… The Mac OS X implementations of a number of routines found in libSystem, including getpwnam and getpwuid, have been modified to be thread-safe. Rather than returning a pointer to a static data structure, these routines allocate a structure for the returned value that is stored in thread-specific memory. The memory is recycled (by each thread) every time you make a call to one of these routines. That allows you to use these APIs safely in multi- threaded applications. The specific routines in the Libinfo project (see Libinfo/ lookup.subproj/lu_*.c) that maintain thread-specific memory for results are: - alias_getbyname alias_getent alias_setent alias_endent - bootp_getbyether bootp_getbyip - bootparams_getbyname bootparams_getent bootparams_setent bootparams_endent - getfsbyname getfsspec getfsfile getfsent setfsent endfsent - initgroups getgrnam getgrgid getgrent setgrent endgrent - gethostbyaddr gethostbyname gethostbyname2 gethostent sethostent endhostent - getipnodebyname getipnodebyaddr - innetgr getnetgrent setnetgrent endnetgrent - getnetbyaddr getnetbyname getnetent setnetent endnetent - prdb_getbyname prdb_get prdb_set prdb_end - getprotobyname getprotobynumber getprotoent setprotoent endprotoent - getrpcbyname getrpcbynumber getrpcent setrpcent endrpcent - getservbyport getservbyname getservent setservent endservent - getpwnam getpwuid getpwent setpwent endpwent Does this also mean when the thread terminates that the memory gets recycled. I could verify that the memory in the same thread gets recycled if I call getpwuid but I still have the memory allocated after my thread terminates. while running my App using "ObjectAlloc" I get memory allocation that are not deallocated / static allocated when calling Looking at the man pages I can not find any reference to the fact that I'm responsible for releasing the memory allocated in getpwuid. Is this a bug in the documentation or is this a bug in the OS that with each call to getpwuid memory gets allocated. This email sent to site_archiver(a)lists.apple.com

1 0

Re: INFINITY and -pedantic with gcc
by Matt Hoffman 03 Aug '05

03 Aug '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com #include <limits> #include <cmath> #include <iostream> using namespace std; int main (int argc, char** argv) { double inf1 = numeric_limits<double>::infinity(); cout << "Infinity: " << inf1 << endl; /* double inf2 = INFINITY; cout << "Infinity: " << inf2 << endl; if (inf1 == inf2) cout << "They're equal." << endl; */ return 0; } -matt On Aug 3, 2005, at 11:59 AM, Jasmin Lapalme wrote: using namespace std; void test_inf(double &x) { x = INFINITY; } int main() { double x; test_inf(x); cout << x << endl; } =========== Thanks Jasmin This email sent to mhoffman(a)cs.washington.edu _______________________________________________ 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 could instead use <limits> and the numeric_limits class. Just a quick example: Un-comment the code involving INFINITY, and it won't compile with - pedantic, but you'll see that they're equivalent. The above code, will however compile with -pedantic. I think this is the better way to do it as well, but that could just be me. Hi, I have curious bug with gcc and the option -pedantic. I want to use the value INFINITY defined in math.h. Here is a sample code : ==== test.cpp ==== #include <cmath> #include <iostream> This code compiles well without the option -pedantic but I want to use the option -pedantic. I want to be sure my code will compile on any platform with gcc. I know that INFINITY is equal to 1e50f and 1e50f is greater than FLT_MAX. The option -pedantic raise this error : floating constant exceeds range of 'float'. Why I cannot use INFINITY with the option -pedantic? _______________________________________________ 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/mhoffman% 40cs.washington.edu This email sent to site_archiver(a)lists.apple.com

1 0

Re: getpwuid memory leak?!?
by Marc Majka 03 Aug '05

03 Aug '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com -- Marc Majka On 3 Aug, 2005, at 10:16, Martin Bestmann wrote: Hi, getpwuid Here is a stack crawl returned by "ObjectAlloc" getpwuid getpw getpw_internal copy_user strdup malloc malloc_zone_malloc getpwuid getpw getpw_internal copy_user calloc I'm running on Mac OS X 10.4.2 Thanks, Martin _______________________________________________ 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/majka%40apple.com This email sent to majka(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… The Mac OS X implementations of a number of routines found in libSystem, including getpwnam and getpwuid, have been modified to be thread-safe. Rather than returning a pointer to a static data structure, these routines allocate a structure for the returned value that is stored in thread-specific memory. The memory is recycled (by each thread) every time you make a call to one of these routines. That allows you to use these APIs safely in multi-threaded applications. The specific routines in the Libinfo project (see Libinfo/ lookup.subproj/lu_*.c) that maintain thread-specific memory for results are: - alias_getbyname alias_getent alias_setent alias_endent - bootp_getbyether bootp_getbyip - bootparams_getbyname bootparams_getent bootparams_setent bootparams_endent - getfsbyname getfsspec getfsfile getfsent setfsent endfsent - initgroups getgrnam getgrgid getgrent setgrent endgrent - gethostbyaddr gethostbyname gethostbyname2 gethostent sethostent endhostent - getipnodebyname getipnodebyaddr - innetgr getnetgrent setnetgrent endnetgrent - getnetbyaddr getnetbyname getnetent setnetent endnetent - prdb_getbyname prdb_get prdb_set prdb_end - getprotobyname getprotobynumber getprotoent setprotoent endprotoent - getrpcbyname getrpcbynumber getrpcent setrpcent endrpcent - getservbyport getservbyname getservent setservent endservent - getpwnam getpwuid getpwent setpwent endpwent while running my App using "ObjectAlloc" I get memory allocation that are not deallocated / static allocated when calling Looking at the man pages I can not find any reference to the fact that I'm responsible for releasing the memory allocated in getpwuid. Is this a bug in the documentation or is this a bug in the OS that with each call to getpwuid memory gets allocated. This email sent to site_archiver(a)lists.apple.com

1 0

Re: INFINITY and -pedantic with gcc
by Markus Hitter 03 Aug '05

03 Aug '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Am 03.08.2005 um 20:59 schrieb Jasmin Lapalme: Behaviour unexpected at the first glance isn't always a bug. Because -pedantic's reason of existence is to catch such oddities. HTH, Markus - - - - - - - - - - - - - - - - - - - Dipl. Ing. Markus Hitter http://www.jump-ing.de/ _______________________________________________ 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 have curious bug with gcc and the option -pedantic. This code compiles well without the option -pedantic but I want to use the option -pedantic. I want to be sure my code will compile on any platform with gcc. I don't think -pedantic helps you here. g++ is g++ on any platform, especially at the front end side. Differences should be limited to optimizer, assembler and linker. I know that INFINITY is equal to 1e50f and 1e50f is greater than FLT_MAX. The option -pedantic raise this error : floating constant exceeds range of 'float'. Raises immediately the question for me: Why would one knowingly want to assign a value greater than FLT_MAX? Anything greater than a Maximum is invalid, by definition. Do you want the positive or the negative infinity, btw.? Why I cannot use INFINITY with the option -pedantic? This email sent to site_archiver(a)lists.apple.com

1 0

Re: valgrind for Darwin or equivalent?
by Craig Rodrigues 03 Aug '05

03 Aug '05

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com User-agent: Mutt/1.5.9i On Mon, Aug 01, 2005 at 07:49:22PM -0700, Eric Albert wrote: > I haven't looked into this since then, so I don't know if the FreeBSD > version has moved forward or been dropped. Doug Rabson ( http://www.rabson.org ) has been working on adding FreeBSD support to valgrind.....we have used it at work, and it works. valgrind is available in the FreeBSD ports collection: http://www.freebsd.org/cgi/cvsweb.cgi/ports/devel/valgrind/ Keep in mind, it is only works for i386 on FreeBSD, but it does work. > Either way, it'd be great > if someone managed to make Valgrind work once the first Intel-based > Macs ship. valgrind relies on some pretty low-level stuff in terms of libc and loading libraries, but hopefully the newer work to get it working on FreeBSD can help if someone wants to port it to MacOS X. -- Craig Rodrigues rodrigc(a)crodrigues.org _______________________________________________ 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

INFINITY and -pedantic with gcc
by Jasmin Lapalme 03 Aug '05

03 Aug '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com using namespace std; void test_inf(double &x) { x = INFINITY; } int main() { double x; test_inf(x); cout << x << endl; } =========== Thanks Jasmin _______________________________________________ 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, I have curious bug with gcc and the option -pedantic. I want to use the value INFINITY defined in math.h. Here is a sample code : ==== test.cpp ==== #include <cmath> #include <iostream> This code compiles well without the option -pedantic but I want to use the option -pedantic. I want to be sure my code will compile on any platform with gcc. I know that INFINITY is equal to 1e50f and 1e50f is greater than FLT_MAX. The option -pedantic raise this error : floating constant exceeds range of 'float'. Why I cannot use INFINITY with the option -pedantic? This email sent to site_archiver(a)lists.apple.com

1 0

Re: Sending generic USB messages?
by seebs＠plethora.net 03 Aug '05

03 Aug '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <FEFBE9B0-F4C6-4FB9-8902-2D279FF4C7F2(a)mac.com>, Robert Palmer Jr wri tes: >Peter, > >There is some sample code on Apple's site for connecting to a USB >device and talking to it. Don't remember the name/section - if you >can't find it, drop me a line privately and I'll dig up the name. >It's pretty simple, only caveat is that the sample code runs in >application space, not driver space, so the USB device has to be >plugged in when the search code executes or it won't find it. It >will let you do the read/write thing - that's what I'm doing. The problem is that the code fragment they give for that seems to work with devices with no drivers. However, someone on the USB list pointed me at setReport and getReport, which appear to do what I want. -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

getpwuid memory leak?!?
by Martin Bestmann 03 Aug '05

03 Aug '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi, getpwuid Here is a stack crawl returned by "ObjectAlloc" getpwuid getpw getpw_internal copy_user strdup malloc malloc_zone_malloc getpwuid getpw getpw_internal copy_user calloc I'm running on Mac OS X 10.4.2 Thanks, Martin _______________________________________________ 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… while running my App using "ObjectAlloc" I get memory allocation that are not deallocated / static allocated when calling Looking at the man pages I can not find any reference to the fact that I'm responsible for releasing the memory allocated in getpwuid. Is this a bug in the documentation or is this a bug in the OS that with each call to getpwuid memory gets allocated. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Sending generic USB messages?
by Robert Palmer Jr 03 Aug '05

03 Aug '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Peter, Robert On Aug 2, 2005, at 6:26 PM, Peter Seebach wrote: While it's true that other OSes treat HID as a generic USB byte stream, this behavior isn't provided for in the USB specification. Apple tries to stick pretty close to the USB spec. This email sent to robertpalmerjr(a)mac.com ----------------------------- Robert G. Palmer, Jr. robertpalmerjr(a)mac.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… There is some sample code on Apple's site for connecting to a USB device and talking to it. Don't remember the name/section - if you can't find it, drop me a line privately and I'll dig up the name. It's pretty simple, only caveat is that the sample code runs in application space, not driver space, so the USB device has to be plugged in when the search code executes or it won't find it. It will let you do the read/write thing - that's what I'm doing. In message <99050708-3021-463C-AEEA-528BC2328E26(a)apple.com>, Garth Cummings wri tes: The problem is that I have a HID device which works in only one way: Send a five-byte packet, receive a 5-byte response. It's not a keyboard or mouse or anything like it, and there's no way to get past this. Anyway, I'll see about the USB lists. It would suck, though, if every other OS I'm looking at targeting let me write the "driver" in 5 minutes on top of the existing HID support, and Darwin made me reinvent the wheel. -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/robertpalmerjr% 40mac.com This email sent to site_archiver(a)lists.apple.com

1 0

Code to walk a process stack?
by Thomas Engelmeier 02 Aug '05

02 Aug '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com MoreIsBetter has some pretty basic code for Mach-O stacktraces: <http://developer.apple.com/samplecode/MoreIsBetter/MoreIsBetter.html> I'm not sure if it also works with plain Darwin calls. HTH, Tom_E _______________________________________________ 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'd like the ability to dynamically know how a certain function was called within a program. That is, I'd >like to be able to walk the stack and see a trace of all the calls. There were some code snippets for >doing this under the old MacOS, but I haven't been able to root one out myself. Can anyone suggest a >place to look? This email sent to site_archiver(a)lists.apple.com

1 0