June 2007 - Darwin-kernel - lists.apple.com

Re: Debugging in kernel
by Terry Lambert 28 Jun '07

28 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi list, printf If you have an XServe, you can also use the serial console, instead. -- Terry _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… On Jun 28, 2007, at 3:32 PM, billy lau wrote: I am currently trying to modify some memory allocators in darwin, for example, the zalloc. However, I am not familiar with the debugging APIs for darwin or Mac OS X, as I am very new to this operating system. Can someone please shed some light on this? Mostly, we use gdb. How to do this is documented in "two machine debugging" and "kernel debugging" at <http://developer.apple.com> in various places (look for the kernel debug kits). For example, is there an equivalent of printk (used in linux) in mac? And if there is, how can I see the result of the these 'prints'? This is covered at <http://developer.apple.com>; typically, you set your boot-args to contain debug=0x14e; that causes the output to go to the console, and you can tail -f /var/log/system.log. Are they printed to some sort of buffer maintained the kernel? Yes, which is then sent to the console device, which is then logged to the system.log. As for linux, printk would basically just print to the console, but, as for Mac, how would that be done, since the operating systems starts in the user graphical interface. If you _can't_ look at the log because you are doing work that causes the system to crash, then you can look at the console ring buffer from kernel space using gdb and two machine debugging (<http://developer.apple.com > again...); alternately, you can login to the non-graphical console, which will cause them to show (more or less) immediately. Note that there is a (very small) latency, so if you are printing e.g. single letters with \n's, and they are very close together source-line-wise, you could lose an event. This is discussed on <http://developer.apple.com>, but basically if you login with the username >console and no password, it will give you a non-graphical login. This email sent to site_archiver(a)lists.apple.com

1 0

Debugging in kernel
by billy lau 28 Jun '07

28 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Dkim-signature: a=rsa-sha1; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:sender:to:subject:mime-version:content-type:x-google-sender-auth; b=qtH3m3E8jvMZBbXu3ul59DhAYcEyQO/6AefmGbX18w8uuDuUgo2oDUcNAikOyAo6ZdgF3qx5PWcipZOzHVAAtb/aj08yHfMK3skpoSf9kQR0jzo1/DWF/O7ko4SXK38V/2OAkIokbD4LqkUxplgVs7ve2Rua8Adcg8Dxf/Ld4u4= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:sender:to:subject:mime-version:content-type:x-google-sender-auth; b=rR+hwKPelTghIirFcF/ZpqbHwUKwMeE5ihbP+roNK+sZ64utUwRC4KHxg1vGqYWcAEcA+FNeqn1D4MJ7Br9f/wKUXq0xQhYNTqxia/8FefGY401HXiBt/rFN0bB3Fb1nWAoU+QDIZ9UI9YWSRyPo2GG+DeMtAWbD9IhmfTAhgzE= _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: enetlognke interface filter
by Michael Smith 27 Jun '07

27 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Jun 27, 2007, at 6:33 AM, Rakesh Singhal wrote: = Mike _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… I am using enetlognke interface filter to print the packets, which are moving on network. For incoming packets, i cant see the packet header, so I dont have information of Destination Addr, Source Addr and Ethernet Type. Can you tell me how to get the whole packet? If this is what you are trying to do, you should move your work to user space. See the manpage for the pcap library: 'man pcap'. This email sent to site_archiver(a)lists.apple.com

1 0

enetlognke interface filter
by Rakesh Singhal 27 Jun '07

27 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Dkim-signature: a=rsa-sha1; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:to:subject:mime-version:content-type; b=rbsznantFPyvFjq21bIu9oIWRtTNhv9x1bWHuoc/Exnl4oJ5Ay2VaM5Cedva6g6iORf0omIF0srexMiVI9NLbVWUV+eWKijZYCE/nF7GlOrfkpNtjA90EqlmpDkBQWSeoB2ZZw5t6ViQyFqEoDMkk5GTra97CSCPD9N6n8JQNx0= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:mime-version:content-type; b=H+Kotzdv8lnJusQVZDnVOB6cWsQAUbgzsuL6iupgkUAiaOINyG3GsStuyiaoVTsHZfnxa32856Bdeh+TniHG9qJT4G9lWwkAyNAslgHAs9XaOEE3VxYjUzgtSctOz80CcUttBMn1EY79uqeGa8DxH/PqPyICwDP+R7zvU/T4pw4= _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: A question of Rosetta.
by Shawn Erickson 25 Jun '07

25 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Dkim-signature: a=rsa-sha1; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=YtnyZvVRtc22MdeuR5wJJt8VtImYXflcOWILy8NkYpRFHtTwEj00HfVIZSgnNGS8CdYonSnWcRyKr/ixQNFkbszzWRlIfmbAfHoZhzKgTtO2II3vNl9w7qKSIm/XBM6dy2ATxPegsdS1WK5Sf/dCd/r3WoKDA0VWVc4mGJMc298= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=U9A0uqIJorie4yiEd2hrMNH6IsecO8Bl7NKi2Kqv+UllMavrB8eEEm39LgximIVd4a1xZ2SvQHanuRdcU9oF/msaIrGmIwcTl3NGKKXAkiy9dgrhCuAr+Acmu2hohKkPMbLHezlqOLOanis+E9b2s75NPDQzY56HSu532gZYVDk= Hi, there, Since Rosetta is a translation process that allows us to use PowerPC-based apps on Intel-based Mac and it's not a simulator of PowerPC, then I suppose it's a translator of machine instructions. If so, why Java applets in a web browser that Rosetta can run translated will not be able to loaded? (Suppose we already have correct JVM installed.) And how about flash plugins? Rosetta doesn't support mixed processes (has to be all PowerPC code) and Java on Intel doesn't run under Rosetta. As a result of those two Java on Intel cannot use PowerPC libraries nor can a Rosetta run browser use Java. <http://developer.apple.com/documentation/MacOSX/Conceptual/universal_binary…> <http://developer.apple.com/qa/qa2005/qa1295.html> -Shawn p.s. please don't cross post _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… On 6/24/07, Calvin (Yu-Hui) Liu <koifans(a)shaw.ca> wrote: This email sent to site_archiver(a)lists.apple.com

1 0

Re: Kernel panic help - kernel controls?
by Greg 23 Jun '07

23 Jun '07

site_archiver(a)lists.apple.com Delivered-To: Darwin-kernel(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Kernel panic help - kernel controls?
by Shawn Erickson 23 Jun '07

23 Jun '07

site_archiver(a)lists.apple.com Delivered-To: Darwin-kernel(a)lists.apple.com On Jun 23, 2007, at 2:31 PM, Greg wrote: #define MAXLEN 64 struct ItemStruct { int len; int flags; char inQueue; #if ALIGNMENT_HACK char *name; #else char name[MAXLEN]; #endif char *other; }; The panic occurs when I try to free it like so: kern_os_free(item->other); kern_os_free(item); -Shawn _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… I was able to fix the panic, the culprit turned out to be some code I was using (seemingly from a trustworthy and reputable source) that recycled memory allocations (to prevent unnecessary mallocing). Forgive me for the misguided accusations at kernel controls, it's just that I did not trust my understanding of them and the backtraces from gdb were fairly vague. I then came across another panic that was caused by what I *think* are boundary alignment problems in a data structure that I have, I've implemented a simple hack-like fix for it but I'm sure there has to be a better way. Sounds like something is writing outside of name[MAXLEN] and overwriting the pointer in other. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Kernel panic help - kernel controls?
by Greg 23 Jun '07

23 Jun '07

site_archiver(a)lists.apple.com Delivered-To: Darwin-kernel(a)lists.apple.com #define MAXLEN 64 struct ItemStruct { int len; int flags; char inQueue; #if ALIGNMENT_HACK char *name; #else char name[MAXLEN]; #endif char *other; }; The panic occurs when I try to free it like so: kern_os_free(item->other); kern_os_free(item); It will not panic if I do not free item->other. kern_os_free(item->name); kern_os_free(item->other); kern_os_free(item); Anyone know why this is happening? - Greg On Jun 22, 2007, at 11:53 PM, Greg wrote: _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… I was able to fix the panic, the culprit turned out to be some code I was using (seemingly from a trustworthy and reputable source) that recycled memory allocations (to prevent unnecessary mallocing). Forgive me for the misguided accusations at kernel controls, it's just that I did not trust my understanding of them and the backtraces from gdb were fairly vague. I then came across another panic that was caused by what I *think* are boundary alignment problems in a data structure that I have, I've implemented a simple hack-like fix for it but I'm sure there has to be a better way. The temporary solution that I've implemented is to simply malloc() name as well and do: I tried to stick __attribute__((aligned(4))) and __attribute__ ((packed)) in various places but I'm not very experienced at using those and all the combinations I could think of still resulted in kernel panics. Sorry for the followup email, but this one is a bit more concise and I've been able to narrow the problem down a bit. I'm able to reproduce the kernel panic by connecting and reconnecting the daemon to the KEXT. Again, it establishes two connection on startup to the KEXT via kernel controls. To simplify things right now only the daemon is sending the KEXT information, the KEXT does not send anything back. The system seems stable on the initial connect, but if I quit the daemon and relaunch it, and repeat this several times, eventually the system (10.4.9 PPC), will panic. There does not seem to be any specific pattern as to what will actually set off the panic, it might happen after I move the mouse, or click on something while the daemon is connected. The most common backtrace that I get is this: 0x002fc368 in OSArray::flushCollection (this=0x1d99c80) at / SourceCache/xnu/xnu-792.17.14/libkern/c++/OSArray.cpp:216 216 /SourceCache/xnu/xnu-792.17.14/libkern/c++/OSArray.cpp: No such file or directory. in /SourceCache/xnu/xnu-792.17.14/libkern/c++/OSArray.cpp Current language: auto; currently c++ (gdb) bt #0 0x002fc368 in OSArray::flushCollection (this=0x1d99c80) at / SourceCache/xnu/xnu-792.17.14/libkern/c++/OSArray.cpp:216 #1 0x002fc1ec in OSArray::free (this=0x1d99c80) at /SourceCache/ xnu/xnu-792.17.14/libkern/c++/OSArray.cpp:160 #2 0x00303bfc in OSSet::free (this=0x1e6c1e0) at /SourceCache/xnu/ xnu-792.17.14/libkern/c++/OSSet.cpp:154 #3 0x002d96f8 in IOMemoryDescriptor::free (this=0x2329800) at / SourceCache/xnu/xnu-792.17.14/iokit/Kernel/IOMemoryDescriptor.cpp:2628 #4 0x002dcf88 in IOBufferMemoryDescriptor::free (this=0x2329800) at /SourceCache/xnu/xnu-792.17.14/iokit/Kernel/ IOBufferMemoryDescriptor.cpp:361 #5 0x007cbcc0 in log_level () #6 0x007bf6cc in log_level () #7 0x002ec0cc in is_io_connect_method_scalarI_structureI (connect=0x1c425c40, index=32522880, input=0x21c6cc4, inputCount=36319744, inputStruct=0x21c6ccc "", inputStructCount=0) at /SourceCache/xnu/xnu-792.17.14/iokit/Kernel/IOUserClient.cpp:2394 #8 0x0008c744 in _Xio_connect_method_scalarI_structureI (InHeadP=0x21c6c9c, OutHeadP=0x226eda8) at device/device_server.c:5128 #9 0x0002921c in ipc_kobject_server (request=0x21c6c00) at / SourceCache/xnu/xnu-792.17.14/osfmk/kern/ipc_kobject.c:309 #10 0x000233f8 in mach_msg_overwrite_trap (args=0x1c425c40) at / SourceCache/xnu/xnu-792.17.14/osfmk/ipc/mach_msg.c:1597 #11 0x000abcac in .L_kernel_syscall_trapcall () #12 0x494f4469 in ?? () Cannot access memory at address 0xbfffd070 Cannot access memory at address 0xbfffd078 What could cause this? The KEXT itself is very simple and written in C. It has the two functions (_start(), _stop()) to be loaded and unloaded from the kernel. Although it does use the IOCreateThread function, it is not a C++ IOKit kernel extension. Again, any help with this is greatly appreciated, I've been banging my head over this problem the past several days and have been unable to make much progress. - Greg _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/greg% 40kinostudios.com This email sent to greg(a)kinostudios.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: Getting vnode from filename?
by Greg 23 Jun '07

23 Jun '07

site_archiver(a)lists.apple.com Delivered-To: Darwin-kernel(a)lists.apple.com On Jun 22, 2007, at 12:44 PM, Michael Smith wrote: No I'm using the vnode scope. - Greg _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… I assume you are referring to the fileop scope? Many of the paths that are passed to this scope are not necessarily viable candidates for lookup; for example the 'from' name for rename is gone by the time the kauth handler runs. 'man cscope' Thanks, I didn't know about this useful command. Unfortunately it wasn't able to find example usage of vnode_lookup in the xnu source but it did find the definition, and while that does tell me a bit more about how to use it, I'm still not sure if it does what I think it does, but I think it's a safe bet that it returns (via pointer arg) the vnode corresponding to the path name. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Kernel panic help - kernel controls?
by Greg 23 Jun '07

23 Jun '07

site_archiver(a)lists.apple.com Delivered-To: Darwin-kernel(a)lists.apple.com - Greg _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… Sorry for the followup email, but this one is a bit more concise and I've been able to narrow the problem down a bit. I'm able to reproduce the kernel panic by connecting and reconnecting the daemon to the KEXT. Again, it establishes two connection on startup to the KEXT via kernel controls. To simplify things right now only the daemon is sending the KEXT information, the KEXT does not send anything back. The system seems stable on the initial connect, but if I quit the daemon and relaunch it, and repeat this several times, eventually the system (10.4.9 PPC), will panic. There does not seem to be any specific pattern as to what will actually set off the panic, it might happen after I move the mouse, or click on something while the daemon is connected. The most common backtrace that I get is this: 0x002fc368 in OSArray::flushCollection (this=0x1d99c80) at / SourceCache/xnu/xnu-792.17.14/libkern/c++/OSArray.cpp:216 216 /SourceCache/xnu/xnu-792.17.14/libkern/c++/OSArray.cpp: No such file or directory. in /SourceCache/xnu/xnu-792.17.14/libkern/c++/OSArray.cpp Current language: auto; currently c++ (gdb) bt #0 0x002fc368 in OSArray::flushCollection (this=0x1d99c80) at / SourceCache/xnu/xnu-792.17.14/libkern/c++/OSArray.cpp:216 #1 0x002fc1ec in OSArray::free (this=0x1d99c80) at /SourceCache/xnu/ xnu-792.17.14/libkern/c++/OSArray.cpp:160 #2 0x00303bfc in OSSet::free (this=0x1e6c1e0) at /SourceCache/xnu/ xnu-792.17.14/libkern/c++/OSSet.cpp:154 #3 0x002d96f8 in IOMemoryDescriptor::free (this=0x2329800) at / SourceCache/xnu/xnu-792.17.14/iokit/Kernel/IOMemoryDescriptor.cpp:2628 #4 0x002dcf88 in IOBufferMemoryDescriptor::free (this=0x2329800) at / SourceCache/xnu/xnu-792.17.14/iokit/Kernel/ IOBufferMemoryDescriptor.cpp:361 #5 0x007cbcc0 in log_level () #6 0x007bf6cc in log_level () #7 0x002ec0cc in is_io_connect_method_scalarI_structureI (connect=0x1c425c40, index=32522880, input=0x21c6cc4, inputCount=36319744, inputStruct=0x21c6ccc "", inputStructCount=0) at /SourceCache/xnu/xnu-792.17.14/iokit/Kernel/IOUserClient.cpp:2394 #8 0x0008c744 in _Xio_connect_method_scalarI_structureI (InHeadP=0x21c6c9c, OutHeadP=0x226eda8) at device/device_server.c:5128 #9 0x0002921c in ipc_kobject_server (request=0x21c6c00) at / SourceCache/xnu/xnu-792.17.14/osfmk/kern/ipc_kobject.c:309 #10 0x000233f8 in mach_msg_overwrite_trap (args=0x1c425c40) at / SourceCache/xnu/xnu-792.17.14/osfmk/ipc/mach_msg.c:1597 #11 0x000abcac in .L_kernel_syscall_trapcall () #12 0x494f4469 in ?? () Cannot access memory at address 0xbfffd070 Cannot access memory at address 0xbfffd078 What could cause this? The KEXT itself is very simple and written in C. It has the two functions (_start(), _stop()) to be loaded and unloaded from the kernel. Although it does use the IOCreateThread function, it is not a C++ IOKit kernel extension. Again, any help with this is greatly appreciated, I've been banging my head over this problem the past several days and have been unable to make much progress. This email sent to site_archiver(a)lists.apple.com

1 0