September 2006 - Darwin-dev

Re: API used to access process heap structure
by Martin Knapp-Cordes 11 Sep '06

11 Sep '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi Michael, Thank you for you detailed explanations. Is there any source of a version of heap that I could look at to help me translate what you are saying? It doesn't appear that the darwin version is published. The man page says it is of BSD origin, but I can't seem to find any trace of it on the freebsd website. I am sure I am not looking in the right place. Any help would be appreciated. Sincerely, Martin Knapp-Cordes --------------------------------------------------------------- | Martin Knapp-Cordes martin(a)mathworks.com | | The MathWorks, Inc. http://www.mathworks.com | | 3 Apple Hill Drive Natick, MA 01760 | | Tel: (508) 647-7321 Fax: (508) 647-7015 | --------------------------------------------------------------- On Fri, 8 Sep 2006, Michael Smith wrote: > > On Sep 8, 2006, at 12:02 PM, darwin-dev-request(a)lists.apple.com wrote: > > > Windows has an API to walk the heap associated with a process. > > There is no POSIX API for UNIX as far as I know. So it will be > > different for each UNIX variant. Darwin has a heap(1) utility which > > clearly uses an API to get the information. You can see some of > > that by > > looking at the output of nm. However, I have been unable to > > find any published documentation about it. I am interested > > for example in extracting information about the blocks on the > > free list. > > There is no published documentation on these interfaces, and they are > not considered API. > > Before we start talking about all this, though, it's important to > understand > that under Darwin, malloc & friends are just wrappers around a plugin > interface that permits multiple instances of multiple allocators to > co-exist. > > In the common case, there's only one allocator, scalable_malloc, but > quite > often there will be several instances of it (as you can see if you > point Heap > at e.g. Safari). > > If you look at <malloc/malloc.h> you will notice that in the > malloc_zone_t > structure there is a pointer to a malloc_introspection_t structure, > which is > defined further down. > > In this structure are a number of useful function vectors. One of these > is the enumerator. > > Now, it's important to understand that in a utility, the enumerator > comes > from the utility's copy of the system library, but if you have set up > the reader > hook to read from another process' address space, you are looking at > data > structures maintained by that process' copy, which may not be the same > (e.g. Rosetta, 32 vs. 64 bit, etc.). > > This particular interface won't tell you what's on the freelist, I'm > afraid, > and you will spend some time digging the zone_address parameter > up if you are introspecting another process (easy enough on your own > task). But it does largely parallel the Windows facility you described, > and if you look at the header it's well enough described that you can > work it out. > > HTH > > = Mike > > _______________________________________________ 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: Disabling IPv4?
by Dan Shoop 09 Sep '06

09 Sep '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Sep 9, 2006, at 11:13 , Dan Shoop wrote: -josh On Sep 7, 2006, at 6:18 PM, Andre-John Mas wrote: Hi, -dhan ------------------------------------------------------------------------ Dan Shoop AIM: iWiring Systems & Networks Architect http://www.ustsvs.com/ shoop(a)iwiring.net http://www.iwiring.net/ 1-714-363-1174 "The wise man doesn't give the right answers, he poses the right questions." -- Claude Levi-Strauss ------------------------------------------------------------------------ iWiring provides systems and networks support for Mac OS X, unix, and Open Source application technologies at affordable rates. _______________________________________________ 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 11:25 AM -0700 9/9/06, Justin C. Walker wrote: At 6:24 PM -0700 9/7/06, Josh Graessley wrote: Content-Type: multipart/signed; micalg=sha1; boundary=Apple-Mail-7--1017169588; protocol="application/pkcs7-signature" If you are running Mac OS X, you can select "Off" for the Configuration method in the Network Preferences pane. I want to do some tests with regards to IPv6 to see how well certain services work on a pure IPv6 environment. Is it possible to disable IPv4 to prevent false positives in my tests? Andre I believe you can turn IPv6 off, but not IPv4. The poster seems to want to disable IPv4, not IPv6. Looks to me like Josh is correct: in the network prefs panel, select the interface in question, then click the TCP/IP tab, and select 'off' in the 'Configure IPv4' popup. IPv6 has a separate configure button, near the bottom. Thanks Justin for pointing that out. I normally like to look before I speak but missed that completely. -- This email sent to site_archiver(a)lists.apple.com

1 0

Re: Disabling IPv4?
by Justin C. Walker 09 Sep '06

09 Sep '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Sep 9, 2006, at 11:13 , Dan Shoop wrote: -josh On Sep 7, 2006, at 6:18 PM, Andre-John Mas wrote: Hi, IPv6 has a separate configure button, near the bottom. Justin -- Justin C. Walker, Curmudgeon at Large Institute for the Absorption of Federal Funds ----------- I'm beginning to like the cut of his jibberish. ----------- _______________________________________________ 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 6:24 PM -0700 9/7/06, Josh Graessley wrote: Content-Type: multipart/signed; micalg=sha1; boundary=Apple- Mail-7--1017169588; protocol="application/pkcs7-signature" If you are running Mac OS X, you can select "Off" for the Configuration method in the Network Preferences pane. I want to do some tests with regards to IPv6 to see how well certain services work on a pure IPv6 environment. Is it possible to disable IPv4 to prevent false positives in my tests? Andre I believe you can turn IPv6 off, but not IPv4. The poster seems to want to disable IPv4, not IPv6. Looks to me like Josh is correct: in the network prefs panel, select the interface in question, then click the TCP/IP tab, and select 'off' in the 'Configure IPv4' popup. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Disabling IPv4?
by Dan Shoop 09 Sep '06

09 Sep '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com -josh On Sep 7, 2006, at 6:18 PM, Andre-John Mas wrote: Hi, -- -dhan ------------------------------------------------------------------------ Dan Shoop AIM: iWiring Systems & Networks Architect http://www.ustsvs.com/ shoop(a)iwiring.net http://www.iwiring.net/ 1-714-363-1174 "The wise man doesn't give the right answers, he poses the right questions." -- Claude Levi-Strauss ------------------------------------------------------------------------ iWiring provides systems and networks support for Mac OS X, unix, and Open Source application technologies at affordable rates. _______________________________________________ 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 6:24 PM -0700 9/7/06, Josh Graessley wrote: Content-Type: multipart/signed; micalg=sha1; boundary=Apple-Mail-7--1017169588; protocol="application/pkcs7-signature" If you are running Mac OS X, you can select "Off" for the Configuration method in the Network Preferences pane. I want to do some tests with regards to IPv6 to see how well certain services work on a pure IPv6 environment. Is it possible to disable IPv4 to prevent false positives in my tests? Andre I believe you can turn IPv6 off, but not IPv4. The poster seems to want to disable IPv4, not IPv6. Have you tired configuring IPv6 manually and configuring the IPv4 info to be bogus? This email sent to site_archiver(a)lists.apple.com

1 0

Re: API used to access process heap structure
by Michael Smith 08 Sep '06

08 Sep '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Sep 8, 2006, at 12:02 PM, darwin-dev-request(a)lists.apple.com wrote: There is no published documentation on these interfaces, and they are not considered API. In this structure are a number of useful function vectors. One of these is the enumerator. HTH = Mike _______________________________________________ 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… Windows has an API to walk the heap associated with a process. There is no POSIX API for UNIX as far as I know. So it will be different for each UNIX variant. Darwin has a heap(1) utility which clearly uses an API to get the information. You can see some of that by looking at the output of nm. However, I have been unable to find any published documentation about it. I am interested for example in extracting information about the blocks on the free list. Before we start talking about all this, though, it's important to understand that under Darwin, malloc & friends are just wrappers around a plugin interface that permits multiple instances of multiple allocators to co-exist. In the common case, there's only one allocator, scalable_malloc, but quite often there will be several instances of it (as you can see if you point Heap at e.g. Safari). If you look at <malloc/malloc.h> you will notice that in the malloc_zone_t structure there is a pointer to a malloc_introspection_t structure, which is defined further down. Now, it's important to understand that in a utility, the enumerator comes from the utility's copy of the system library, but if you have set up the reader hook to read from another process' address space, you are looking at data structures maintained by that process' copy, which may not be the same (e.g. Rosetta, 32 vs. 64 bit, etc.). This particular interface won't tell you what's on the freelist, I'm afraid, and you will spend some time digging the zone_address parameter up if you are introspecting another process (easy enough on your own task). But it does largely parallel the Windows facility you described, and if you look at the header it's well enough described that you can work it out. This email sent to site_archiver(a)lists.apple.com

1 0

CFPreferences and non-root daemon
by Marc Epard 08 Sep '06

08 Sep '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AcbTYcLKAYWTOD9VEdu63QANk21umg== Thread-topic: CFPreferences and non-root daemon User-agent: Microsoft-Entourage/11.2.5.060620 I'm moving root-requiring code out of a launchd-started daemon into a couple small, focused executables. It's going well except I've hit a snag with CFPreferences. I'm specifying kCFPreferencesCurrentApplication, KCFPreferencesAnyUser, KCFPreferencesCurrentHost and CFPreferences works fine if I run the daemon as root. But if I run it as nobody, CFPreferencesSynchronize consistently fails. If I manually change the owner of the bundle.id.plist file that CFPreferences creates to nobody, it works okay. None of this is a particular surprise, but I'm wondering what to do about it. How should a non-root daemon store its preferences? The daemon can change its settings during execution, so a static config file is not an option. Is there a good way to get CFPreferences to work? -Marc _______________________________________________ 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: API used to access process heap structure
by Martin Knapp-Cordes 08 Sep '06

08 Sep '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi, Yes, I only meant the current process. I hope that clarifies things. Sincerely, Martin Knapp-Cordes --------------------------------------------------------------- | Martin Knapp-Cordes martin(a)mathworks.com | | The MathWorks, Inc. http://www.mathworks.com | | 3 Apple Hill Drive Natick, MA 01760 | | Tel: (508) 647-7321 Fax: (508) 647-7015 | --------------------------------------------------------------- On Fri, 8 Sep 2006, Mailing list subscriptions wrote: > El 08/09/2006, a las 1:35, Stewart.Dickson(a)noaa.gov escribi?: > > > walking another process's memory heap smacks of > > The original poster never said anything about examining the memory of > another process. He may have meant that he wanted to examine the > memory of the current process. > > > _______________________________________________ > 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/martin%40mathworks.com > > This email sent to martin(a)mathworks.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: API used to access process heap structure
by Mailing list subscriptions 08 Sep '06

08 Sep '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com El 08/09/2006, a las 1:35, Stewart.Dickson(a)noaa.gov escribió: _______________________________________________ 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… walking another process's memory heap smacks of The original poster never said anything about examining the memory of another process. He may have meant that he wanted to examine the memory of the current process. This email sent to site_archiver(a)lists.apple.com

1 0

Re: API used to access process heap structure
by Pelle Johansson 08 Sep '06

08 Sep '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com 8 sep 2006 kl. 01.35 skrev Stewart.Dickson(a)noaa.gov: walking another process's memory heap smacks of 1. reverse engineering Can't really argue here, it could be used for this. or 3. trying to patch a memory-leak which is better done using a debugger like Purify http://en.wikipedia.org/wiki/Purify Much easier done in a normal debugger, or by simply disassembling. It's a perfectly legal reason, btw. 2. developing a buffer-overflow attack or 4. Cheating in games: http://www.brokenzipper.com/. Source code is available, which should help the OP (see especially VMRegion.m). It uses task_for_pid() though, so something has to be done to get around that on Intel. -- Pelle Johansson _______________________________________________ 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: Disabling IPv4?
by Josh Graessley 08 Sep '06

08 Sep '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com -josh On Sep 7, 2006, at 6:18 PM, Andre-John Mas wrote: 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… This email sent to site_archiver(a)lists.apple.com If you are running Mac OS X, you can select "Off" for the Configuration method in the Network Preferences pane. I want to do some tests with regards to IPv6 to see how well certain services work on a pure IPv6 environment. Is it possible to disable IPv4 to prevent false positives in my tests? _______________________________________________ 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/jgraessley% 40apple.com This email sent to jgraessley(a)apple.com smime.p7s

1 0