April 2008 - Darwin-dev - lists.apple.com

Re: failmalloc equivalent for MacOS X
by Army Research Lab 10 Apr '08

10 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AcibDvJ6MTBdJAcCEd2qDwAdT0T19A== Thread-topic: failmalloc equivalent for MacOS X User-agent: Microsoft-Entourage/11.4.0.080122 On 4/10/08 8:49 AM, "David Geldreich" <david.geldreich(a)free.fr> wrote: > Hi all, > > Le 10 avr. 08 à 14:06, Army Research Lab a écrit : >> >> I'd love to have a copy! However, I really think that it should >> also be >> built into libgmalloc. As you noted, and as mentioned on the >> failmalloc >> page, too many people assume that their code will work correctly, >> 100% of >> the time. Even Apple is doing it! :) I'd rather test, test, test, >> and then >> test some more, so that my users don't get stuck with my bad code. >> Memory >> testing should be put into libgmalloc for that reason. >> >> So, I'm willing to work on the source for libgmalloc, if it is >> available for >> download. Is it? > > I have (minimally) packaged my sources and they are available at : > http://david.geldreich.free.fr/dev.html > > I have used it on Tiger but it should work on Leopard as well. > > I am just replacing malloc and realloc and this seems to be enough to > make : > > - malloc/realloc returns NULL > - new throw std::bad_alloc > -NXZone*alloc to fail > > Any bug reports, enhancement, suggestions are welcomed. Thanks, I looked at it, and it looks very clean and simple. I have a couple of ideas of what I'd like to do, I'll think more about them tonight and send you the patches. Thanks, Cem Karan _______________________________________________ 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: failmalloc equivalent for MacOS X
by David Geldreich 10 Apr '08

10 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi all, I have (minimally) packaged my sources and they are available at : http://david.geldreich.free.fr/dev.html I have used it on Tiger but it should work on Leopard as well. - malloc/realloc returns NULL - new throw std::bad_alloc -NXZone*alloc to fail Any bug reports, enhancement, suggestions are welcomed. _______________________________________________ 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… Le 10 avr. 08 à 14:06, Army Research Lab a écrit : I'd love to have a copy! However, I really think that it should also be built into libgmalloc. As you noted, and as mentioned on the failmalloc page, too many people assume that their code will work correctly, 100% of the time. Even Apple is doing it! :) I'd rather test, test, test, and then test some more, so that my users don't get stuck with my bad code. Memory testing should be put into libgmalloc for that reason. So, I'm willing to work on the source for libgmalloc, if it is available for download. Is it? I am just replacing malloc and realloc and this seems to be enough to make : This email sent to site_archiver(a)lists.apple.com

1 0

Re: failmalloc equivalent for MacOS X
by Army Research Lab 10 Apr '08

10 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AcibAz2OfC4IYQb2Ed2qDwAdT0T19A== Thread-topic: failmalloc equivalent for MacOS X User-agent: Microsoft-Entourage/11.4.0.080122 On 4/10/08 7:54 AM, "David Geldreich" <david.geldreich(a)free.fr> wrote: > Hello Cem, > > Le 10 avr. 08 à 13:43, Army Research Lab a écrit : > >> I noticed the message below from David, and it occurred to me that >> this >> might be something to build into libgmalloc. I looked around on >> Apple's >> site for the sources, and then out on the web in general, but I >> didn't find >> them; are they Apple proprietary, or am I just blind? > > I could release my libfailmalloc ... but even I get my application > resist against allocation failure, Apple's frameworks (OpenGL, > QuickTime, CoreVideo, ...) are not "memory allocation problem" safe. > > I stream video decoded with QuickTime to the GPU and I have crash in > QuickTime codec or in OpenGL driver (texture management part)... > > So for my application, the only solution is to not come close the 2GB > limit (for a 32 bit app). If I detect an allocation failure, I > immediately try to free as much memory as possible but it does not > prevent crash. > > If someone wants to test my libfailmalloc, just ask me for a copy. > > Best regards. I'd love to have a copy! However, I really think that it should also be built into libgmalloc. As you noted, and as mentioned on the failmalloc page, too many people assume that their code will work correctly, 100% of the time. Even Apple is doing it! :) I'd rather test, test, test, and then test some more, so that my users don't get stuck with my bad code. Memory testing should be put into libgmalloc for that reason. So, I'm willing to work on the source for libgmalloc, if it is available for download. Is it? Thanks, Cem Karan _______________________________________________ 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: failmalloc equivalent for MacOS X
by David Geldreich 10 Apr '08

10 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hello Cem, Le 10 avr. 08 à 13:43, Army Research Lab a écrit : If someone wants to test my libfailmalloc, just ask me for a copy. 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… I noticed the message below from David, and it occurred to me that this might be something to build into libgmalloc. I looked around on Apple's site for the sources, and then out on the web in general, but I didn't find them; are they Apple proprietary, or am I just blind? I could release my libfailmalloc ... but even I get my application resist against allocation failure, Apple's frameworks (OpenGL, QuickTime, CoreVideo, ...) are not "memory allocation problem" safe. I stream video decoded with QuickTime to the GPU and I have crash in QuickTime codec or in OpenGL driver (texture management part)... So for my application, the only solution is to not come close the 2GB limit (for a 32 bit app). If I detect an allocation failure, I immediately try to free as much memory as possible but it does not prevent crash. This email sent to site_archiver(a)lists.apple.com

1 0

Re: failmalloc equivalent for MacOS X
by Army Research Lab 10 Apr '08

10 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AcibAAjMR6GMXAbzEd2qDwAdT0T19A== Thread-topic: failmalloc equivalent for MacOS X User-agent: Microsoft-Entourage/11.4.0.080122 I noticed the message below from David, and it occurred to me that this might be something to build into libgmalloc. I looked around on Apple's site for the sources, and then out on the web in general, but I didn't find them; are they Apple proprietary, or am I just blind? Thanks, Cem Karan ------------------------------ Message: 4 Date: Thu, 20 Mar 2008 14:11:04 +0100 From: David Geldreich <david.geldreich(a)free.fr> Subject: failmalloc equivalent for MacOS X To: darwin-dev(a)lists.apple.com Message-ID: <D8356345-8B1C-4520-9DBD-8779C7DDF56D(a)free.fr> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Hi everyone, I am trying to test a program against malloc/new failure. I was wondering if an equivalent of failmalloc (http://www.nongnu.org/ failmalloc/) exists for OSX ? failmalloc is dependent on glibc memory allocation scheme, so does not work on OSX. If not, I will write one such library. I am wondering if it is OK to introspect libgmalloc for finding all the malloc related functions that are rerouted for testing ? I need to override less functions (only the one that allocates memory and could fail) but I think libgmalloc have an exhaustive list of entry points. ab5fa114 s __interpose_NXAddRegion ab5fa1fc s __interpose_NXCreateChildZone ab5fa204 s __interpose_NXCreateZone ab5fa20c s __interpose_NXDefaultMallocZone ab5fa214 s __interpose_NXDestroyZone ab5fa1cc s __interpose_NXMallocCheck ab5fa1f4 s __interpose_NXMergeZone ab5fa1c4 s __interpose_NXNameZone ab5fa1d4 s __interpose_NXProtectZone ab5fa1bc s __interpose_NXRemoveRegion ab5fa1ec s __interpose_NXZoneCalloc ab5fa21c s __interpose_NXZoneFree ab5fa1e4 s __interpose_NXZoneFromPtr ab5fa224 s __interpose_NXZoneMalloc ab5fa1dc s __interpose_NXZonePtrInfo ab5fa22c s __interpose_NXZoneRealloc ab5fa1b4 s __interpose__NXMallocDumpFrees ab5fa1ac s __interpose__NXMallocDumpZones ab5fa184 s __interpose__malloc_fork_child ab5fa18c s __interpose__malloc_fork_parent ab5fa194 s __interpose__malloc_fork_prepare ab5fa144 s __interpose_calloc ab5fa154 s __interpose_free ab5fa15c s __interpose_malloc ab5fa10c s __interpose_malloc_create_zone ab5fa174 s __interpose_malloc_debug ab5fa104 s __interpose_malloc_default_zone ab5fa0fc s __interpose_malloc_destroy_zone ab5fa12c s __interpose_malloc_error ab5fa16c s __interpose_malloc_freezedry ab5fa0f4 s __interpose_malloc_get_all_zones ab5fa0ec s __interpose_malloc_get_zone_name ab5fa134 s __interpose_malloc_good_size ab5fa164 s __interpose_malloc_jumpstart ab5fa234 s __interpose_malloc_printf ab5fa0e4 s __interpose_malloc_set_zone_name ab5fa19c s __interpose_malloc_singlethreaded ab5fa13c s __interpose_malloc_size ab5fa0dc s __interpose_malloc_zone_calloc ab5fa0d4 s __interpose_malloc_zone_check ab5fa0cc s __interpose_malloc_zone_free ab5fa0c4 s __interpose_malloc_zone_from_ptr ab5fa0bc s __interpose_malloc_zone_log ab5fa0b4 s __interpose_malloc_zone_malloc ab5fa0ac s __interpose_malloc_zone_print ab5fa0a4 s __interpose_malloc_zone_print_ptr_info ab5fa09c s __interpose_malloc_zone_realloc ab5fa094 s __interpose_malloc_zone_register ab5fa08c s __interpose_malloc_zone_unregister ab5fa124 s __interpose_malloc_zone_valloc ab5fa14c s __interpose_realloc ab5fa1a4 s __interpose_set_malloc_singlethreaded ab5fa11c s __interpose_valloc ab5fa17c s __interpose_vfree Thanks in advance. _______________________________________________ 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

Sources Posted
by William Siegrist 09 Apr '08

09 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com The darwin team has posted sources for the iPhone SDK Beta 3. <http://www.opensource.apple.com/darwinsource/iPhoneSDKBeta3/> Thank you. ---- William Siegrist Darwin Team _______________________________________________ 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 smime.p7s

1 0

Firewall in System preferences & IPv6?
by Andre-John Mas 08 Apr '08

08 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi, 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… Does anyone know whether the firewall configuration in the system preferences, under MacOS X 10.5 is IPv6 aware? This email sent to site_archiver(a)lists.apple.com

1 0

Re: Sources Posted
by Jonas Maebe 08 Apr '08

08 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com SDK Beta: <http://www.opensource.apple.com/darwinsource/iPhoneSDKBeta2/> cctools gcc gdb Anyway, thanks a lot for posting the sources! 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… On 08 Apr 2008, at 04:35, William Siegrist wrote: This is strange: the gdb version for Xcode 3.1 beta (aka iPhone SDK Beta 1) is gdb-908, while the one in Beta 2 is gdb-870. This email sent to site_archiver(a)lists.apple.com

1 0

Sources Posted
by William Siegrist 08 Apr '08

08 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com SDK Beta: <http://www.opensource.apple.com/darwinsource/iPhoneSDKBeta2/> cctools gcc gdb gnumake iPhone 1.1.3 & 1.1.4: <http://www.opensource.apple.com/darwinsource/iPhone/> JavaScriptCore WebCore Thank you. ---- William Siegrist Darwin Team _______________________________________________ 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 The darwin team has posted the sources for the following iPhone- related projects: smime.p7s

1 0

Re: Seek to end?
by Terry Lambert 07 Apr '08

07 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Please file a bug report at <http://bugreport.apple.com>. -- Terry _______________________________________________ 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 Apr 5, 2008, at 7:12 PM, Nathaniel Gray wrote: Peter Seebach writes: Prototypes. Compile with warnings on; you'd have gotten the answer from the compiler. Ah, thanks! No cookie for the genius who decided that "Treat Missing Function Prototypes as Errors" should be *disabled* in Xcode by default... This email sent to site_archiver(a)lists.apple.com

1 0