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

Re: a Bootstrap Namespace question
by Jeremy Pereira 03 Jun '07

03 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On 1 Jun 2007, at 08:18, Serge Cohen 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.… Thanks for the information, it is now much clearer why we did not achieve anything using this route.... If the SUID shell scripts are deactivated, I guess it is for a proper reason, and I'd rather stay with that default settings. Is this overly cautious to keep them deactivated ? No, it's not overly cautious. If you have a root setuid shell script, it is almost trivially easy to get root access. I imagine you could, for instance, put a directory in your home directory in your path ahead of /bin and /usr/bin, put a copy of the shell in that directory but give it the same name as some common command that's in the setuid script and then execute the script. When the script hits that command, it will instead execute your renamed shell - as root. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Reload VM hint? (James Bucanek)
by Sam Vaughan 03 Jun '07

03 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On 01/06/2007, at 5:04 AM, James Bucanek <subscriber(a)gloaming.com>: I hadn't thought of that, and that's an excellent suggestion (thanks also extended to Brian and Matt). The source file data is only read once, so at least I can eliminate that from being cashed. Come to think of it, that might improve my overall performance since the reading the source file blocks also competes with the caching of my database file blocks. _______________________________________________ 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 There's an even better reason why F_NOCACHE will probably improve your code's performance. Provided that you do as Quinn suggests with respect to alignment, it will cause the I/O to go directly to/from your user space buffer, removing the additional copy that happens in the buffered I/O case. This is why it's called "direct I/O" on most other platforms, where the O_DIRECT flag is used at file open time to enable it. Sam smime.p7s

1 0

Re: kdb with serial port
by Terry Lambert 02 Jun '07

02 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thank you Terry for this great infos. -- 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 2, 2007, at 8:14 AM, apic | puredarwin.org wrote: I put right after the delay you mentioned a PE_enter_debugger("debug_me") which works fine, the problem at this point now is the kernel does not find a mac/ip address to use. Which nic's mac address is commonly used? The primary interfaces one? Yes. If this is PPC, and you have bootp setup, then it will set the IP address to the bootp address in Open Firmware, and then once it's done that, even though it says it's IP is 0.0.0.0 when it's waiting for the debugger to attach, the IP address will actually be what the bootp/dhcpd vended to OF. For Intel, I just always use the firewire debugging method (or hardware debugger, if it's too early for IOKit to to have started), which has the same effect. I don't know if Intel sets the IP in EFI or not (maybe Derek is listening in on this and can answer). The way to know if something is registered to handle debugging that early, BTW, is whether or not it says "waiting for debugger" or it says "We are hanging here...". If it says the latter, no one has registered to handle kdp, and you will need to find some way to get a driver loaded (generally, firewire is the recommended metho - see the kernel debugging documentation for details). This email sent to site_archiver(a)lists.apple.com

1 0

Re: kdb with serial port
by apic | puredarwin.org 02 Jun '07

02 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Thank you Terry for this great infos. Am 02.06.2007 um 05:00 schrieb Terry Lambert: The kernel debugging that you are trying to use is based on: (1) having a polled mode driver (2) being able to talk to that polled mode driver with a gdb -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (Darwin) iD8DBQFGYYlaBDxoGUAtYxoRAt8fAKCsNnz81S7ejSOgHFpOZqhCXh+IUwCgn2hl KXXNUSvVZFzOKF7xicsUoqQ= =g86h -----END PGP SIGNATURE----- _______________________________________________ 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 put right after the delay you mentioned a PE_enter_debugger ("debug_me") which works fine, the problem at this point now is the kernel does not find a mac/ip address to use. Which nic's mac address is commonly used? The primary interfaces one? On Jun 1, 2007, at 3:34 PM, apic | puredarwin.org wrote: Thank you for your answer. I now can successfully have the kernel break at the desired point but now comes a further problem into play. I cannot make the kdg to jump on the train of a serial port, also I cannot figure how one would attach to a remote-kdg target via the serial port. This case is more problematic than usale as I have no chance to load any network driver and do debugging via a network connection of any kind. The docs previously posted mention there is this possibility but I can hardly find any docuemnt discribing the serial line approach. This is not supported. The serial debugger is not a gdb based debugger, and it only works if you build your kernel DEBUG and set flags. There's no way to symbolically debug via serial port. Theoretically, you could hack up gdb to support this, and then you could hack up a serial driver that registered itself as the driver to use when you break to the debugger, but then you are in exactly the same boat as before, where you need to load the KEXT. What you need to do is follow the instructions (in the kernel debugging tech notes) so that BootX loads the KEXT for you, by telling it the KEXT is necessary to mount the root device in safe mode; this will get the debugger KEXT loaded early, but you will still need to get it to register itself before you can debug the kernel that way. This basically means that the earliest you can reasonably expect to get to a gdb is in bsd_init(), just prior to it calling inittodr() - and that's exactly where you will see where we, when we are working on the kernel, start our early debugging (there is a small "#if 0" section that you change to a "#if 1" to enable early debugging). Other than that, you are stuck with the built-in serial debugger, unless you have hardware like a JTAG or ITP. Am 01.06.2007 um 22:00 schrieb Herb Petschauer: Put a PE_enter_debugger() call into your start() method? This is frequently not enough, since you have to wait for things to settle to ensure the debug transport device is registered, etc., before jumping into the debugger. That's why the commented out block in bsd_init() has an IOSleep() for 10 seconds. -- Terry This email sent to site_archiver(a)lists.apple.com

1 0

Re: kdb with serial port
by Terry Lambert 02 Jun '07

02 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com The kernel debugging that you are trying to use is based on: (1) having a polled mode driver (2) being able to talk to that polled mode driver with a gdb -- 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 1, 2007, at 3:34 PM, apic | puredarwin.org wrote: Thank you for your answer. I now can successfully have the kernel break at the desired point but now comes a further problem into play. I cannot make the kdg to jump on the train of a serial port, also I cannot figure how one would attach to a remote-kdg target via the serial port. This case is more problematic than usale as I have no chance to load any network driver and do debugging via a network connection of any kind. The docs previously posted mention there is this possibility but I can hardly find any docuemnt discribing the serial line approach. This is not supported. The serial debugger is not a gdb based debugger, and it only works if you build your kernel DEBUG and set flags. There's no way to symbolically debug via serial port. Theoretically, you could hack up gdb to support this, and then you could hack up a serial driver that registered itself as the driver to use when you break to the debugger, but then you are in exactly the same boat as before, where you need to load the KEXT. What you need to do is follow the instructions (in the kernel debugging tech notes) so that BootX loads the KEXT for you, by telling it the KEXT is necessary to mount the root device in safe mode; this will get the debugger KEXT loaded early, but you will still need to get it to register itself before you can debug the kernel that way. This basically means that the earliest you can reasonably expect to get to a gdb is in bsd_init(), just prior to it calling inittodr() - and that's exactly where you will see where we, when we are working on the kernel, start our early debugging (there is a small "#if 0" section that you change to a "#if 1" to enable early debugging). Other than that, you are stuck with the built-in serial debugger, unless you have hardware like a JTAG or ITP. Am 01.06.2007 um 22:00 schrieb Herb Petschauer: Put a PE_enter_debugger() call into your start() method? This is frequently not enough, since you have to wait for things to settle to ensure the debug transport device is registered, etc., before jumping into the debugger. That's why the commented out block in bsd_init() has an IOSleep() for 10 seconds. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Modify CPU registers
by Terry Lambert 02 Jun '07

02 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Yes. I suggest: <http://www.power.org/resources/downloads/PowerISA_Public.pdf> for more reading on what's privileged or not privileged. You also need: -- 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 1, 2007, at 11:54 AM, Raul Flores wrote: I would like to lock/invalidate/flush cache areas and or set other cpu registers on a 7447A processor (Apple G4 PowerBook) with OS X (Darwin) 10.4. Not sure how to proceed, as many of the commands are "supervisor mode". A simple example would be reading HID0 and HID1 registers. Trying to perform an "mfspr" command for any other than 1,8,or 9 results in an illegal exception, when executed from the user state. Any ideas on how to approach/implement this type of functionality from my software? It appears that the CHUD framework installs a KEXT to control the performance counter registers. Am I looking at the same type of implementation (KEXT) to have my user application query/modify some of the register settings? Control registers require that you be running in supervisor state to examine or modify. <http://www.freescale.com/files/microcontrollers/doc/user_guide/DEMO9RS08KA2… > But I'm told on good authority that you will find that whatever you change, we will likely either crash or change it back to what it's supposed to be, instead of what you set it to. This email sent to site_archiver(a)lists.apple.com

1 0

Re: kdb with serial port
by apic | puredarwin.org 01 Jun '07

01 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I tried these flags so far: 0x1,0x11 = dos nothing, does not break 0x1F,0x11F = breaks wants network 0xF = same as above, at least gives output to serial port Any ideas? This serial line thing, drives me nuts already. Thanks to anyone for help. Am 01.06.2007 um 22:00 schrieb Herb Petschauer: -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (Darwin) iD8DBQFGYJ8GBDxoGUAtYxoRAsVNAKCHFenx09//8dB7dyMnDYFXyLPZ9wCgkkmK vcVc1LNPY0vhcBvdSvl9UyY= =J5EB -----END PGP SIGNATURE----- _______________________________________________ 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.… Thank you for your answer. I now can successfully have the kernel break at the desired point but now comes a further problem into play. I cannot make the kdg to jump on the train of a serial port, also I cannot figure how one would attach to a remote-kdg target via the serial port. This case is more problematic than usale as I have no chance to load any network driver and do debugging via a network connection of any kind. The docs previously posted mention there is this possibility but I can hardly find any docuemnt discribing the serial line approach. Put a PE_enter_debugger() call into your start() method? This email sent to site_archiver(a)lists.apple.com

1 0

Re: Problem building darwin kernel
by apic | puredarwin.org 01 Jun '07

01 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 iD8DBQFGYH+lBDxoGUAtYxoRAjfoAJ9/ZkpkJyDXTLrJ8+dMww1i6PSH1gCfepXq 1XjKHqczIbT/sjsZupyyh/I= =fjQG -----END PGP SIGNATURE----- _______________________________________________ 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.… [ building mach_kernel ] /bin/sh: line 1: /Users/blau2/xnu-792.17.14/config/newvers.pl: Permission denied make[2]: *** [do_build_mach_kernel] Error 1 make[1]: *** [build_all] Error 2 make: *** [all] Error 2 Build as root (sudo -s) or change permissions of files to the user compiling. :) -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (Darwin) This email sent to site_archiver(a)lists.apple.com

1 0

Re: Problem building darwin kernel
by Rick Macklem 01 Jun '07

01 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Fri, 1 Jun 2007, Billy Lau wrote: Thanks for the help given so far. Say now that I assume that the tools has been properly installed, and I proceed all the way to compile the kernel, and by the way, this is done on a powerPC. As I was executing "make exporthdrs", everything went well, however, when I execute "make all", it proceeded all the way until it gave me this: [ building mach_kernel ] /bin/sh: line 1: /Users/blau2/xnu-792.17.14/config/newvers.pl: Permission denied make[2]: *** [do_build_mach_kernel] Error 1 make[1]: *** [build_all] Error 2 make: *** [all] Error 2 What possibly went wrong? And by the way, I am very new to this development environment. I really don't know much about what does errors stand for, and how do you guys check for interdependencies and so on. That file is in the xnu-792.17.14 tarball, so my only thoughts are: 1 - "make exporthdrs" deleted it for some reason. I don't bother doing "make exporthdrs" when I build a kernel, just "make" when in xnu-792.17.14. OR 2 - You just didn't unroll the tarball correctly? rick _______________________________________________ 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

missing headerfile in compilation
by Billy Lau 01 Jun '07

01 Jun '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com As I am trying to compile the kextsymboltool.c, the compiler tells me: kextsymboltool.c:21:24: mach-o/kld.h: No such file or directory In file included from /System/Library/Frameworks/IOKit.framework/PrivateHeaders/kext/KXKextManager.h:113, from kextsymboltool.c:28: /System/Library/Frameworks/IOKit.framework/PrivateHeaders/kext/KXKext.h:10:24: mach-o/kld.h: No such file or directory But previously, I had already done the following: sudo mkdir -p /System/Library/Frameworks/IOKit.framework/Versions/A/PrivateHeaders/kext cd /System/Library/Frameworks/IOKit.framework/ sudo ln -s Versions/A/PrivateHeaders PrivateHeaders sudo cp PATH_TO_IOKITUSER/IOKitUser-version/kext.subproj/*.h PrivateHeaders/kext Any help? Thanks _______________________________________________ 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