July 2016 - Darwin-kernel - lists.apple.com

RE: Problem with dlopen
by Emmanuel T. 25 Jul '16

25 Jul '16

_______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Problem with dlopen
by Gordo Cath 25 Jul '16

25 Jul '16

Hi, I'm using 10.10.5 and using the code from http://www.newosxbook.com/src.jl?tree=listings&file=inject.c I created my own simple library from the sample at the bottom of the file using Xcode 6.2. everything compiles fine and I'm not arm64. When I go to run i use 2 instances of lldb. one on the target and one in the one running inject. I setup a breakpoint on dlopen in the target. I watch it go through the process of loading my library. of interest is rbx appears to have the name of the library and r14 is set to 0. I have seen other runtime examples have r14 set to name of the library and rbx set to the number of characters of the name of the library. Anyways, it always crashes at addr 0x7fff62ac56097 0f 29 45 c0 movaps %xmm0 -0x40(:rbp) -- which if i understand it is trying to save off a value to a location 0x40 ahead of where base pointer is now at. I've checked my settings and it appears to be a legitimate location. The values are all 0 The instruction right before it is a xorps %xmm0, %xmm0 I don't know if the 0x7fff62ac56097 holds across other programs but it /usr/lib/dyld.dylib I tried googling to see if this is a known bug. I'm wondering if not having the # characters set in the name in a register is the problem. any ideas? Thanks Gordo _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: posix_spawn() and the KAUTH_FILEOP_EXEC action (in KAUTHs' 'FILEOP' scope)
by Phil Dennis-Jordan 20 Jul '16

20 Jul '16

_______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: posix_spawn() and the KAUTH_FILEOP_EXEC action (in KAUTHs' 'FILEOP' scope)
by Luigi Vampa 19 Jul '16

19 Jul '16

_______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: posix_spawn() and the KAUTH_FILEOP_EXEC action (in KAUTHs' 'FILEOP' scope)
by Phil Dennis-Jordan 19 Jul '16

19 Jul '16

_______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

posix_spawn() and the KAUTH_FILEOP_EXEC action (in KAUTHs' 'FILEOP' scope)
by Luigi Vampa 19 Jul '16

19 Jul '16

_______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Regd. tracking 'child pid' during 'fork'
by Charu Tiwari 15 Jul '16

15 Jul '16

_______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Getting a task_t via PID from IOKit driver
by Nicoló Blasi 08 Jul '16

08 Jul '16

Hi, is there a (somewhat) clean way to get a task_t structure for a userland task from its PID, from an IOKit driver? I just need a way to access a userland task from my userclient, so I can manipulate that task address space from kernel. I know this is a terrible idea and should not be done in any case, but I am just experimenting and do not plan to release this driver publicly for any serious purpose. task_for_pid mach trap is not available for use, from what I have seen. _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: I'd like to be able to set registers directly
by William Kucharski 07 Jul '16

07 Jul '16

> On Jul 7, 2016, at 7:25 AM, Gordo Cath <ds(a)y23.org> wrote: > > Hi William, > Thanks for the help. I'm guessing the register keyword doesn't address variables directly or doesn't do what I thought it was doing. You are correct that you are incorrect. :-) The "register" keyword is a largely deprecated "hint" to the compiler that it may want to store that variable in a register for faster access; in reality most C optimizers are pretty good at figuring out what will go in registers and what wont, so the register keyword is largely if not completely ignored. To actually use registers you have to use assembly instructions and the extended asm syntax of the form I gave a sample of. William Kucharski _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0

Re: I'd like to be able to set registers directly
by Gordo Cath 07 Jul '16

07 Jul '16

Hi William, Thanks for the help. I'm guessing the register keyword doesn't address variables directly or doesn't do what I thought it was doing. I'm a bit new to apple. What I've discovered is in my thread is I have no standard I/O where I'm crashing so I have no print or logging capability. Thus I'm now leaning towards using unused registers to figure out where its crashing. I also don't have a good way to use a debugger afaik unless I could trigger the debugger at crash time. it used to do so but for some reason it isn't doing it now. I have the nvram settings setup to do so but its not triggering. gordo > -------Original Message------- > From: William Kucharski <kucharsk(a)mac.com> > To: Gordo Cath <ds(a)y23.org> > Cc: darwin-kernel(a)lists.apple.com > Subject: Re: I'd like to be able to set registers directly > Sent: Jul 07 '16 08:56 > > > > On Jul 7, 2016, at 6:36 AM, Gordo Cath <ds(a)y23.org> wrote: > > > > Hi, > > I'd like to be able to set a register directly from my c code. I'm using the llvm compiler in a x86_64 build/run environment. > > > > When I compile and link the following code I get no errors or warnings: > > > > I used the thread_create_running function and during setup set the r11 register to 0xBEAD as a start state using x86_thread_state64_t . > > > > ///snippet---- at the very start of my thread I have the following tidbit of code: > > register uint64_t r11 asm("r11"); > > r11=0xfade; // some value -- don't care about the value, just want my value set. > > //end snippet > > What, precisely, are you trying to do? As far as I know, that asm syntax is just wrong. > > I would expect to see something like this for x86_64: > > #include <stdio.h> > #include <sys/types.h> > > int > main() > { > int64_t r11; > > r11 = 0xc01ddeadbeef; /* init r11 */ > > asm ("movq %0, %%r11" :: "r"(r11) : "r11"); /* move r11 to %r11 */ > r11 = 0; /* zero r11 */ > asm ("movq %%r11, %0" : "=r"(r11) ::); /* move %r11 to r11 */ > > (void) printf("%llx\n", r11); /* verify %r11 */ > return 0; > } > > > William Kucharski _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists… This email sent to site_archiver(a)lists.apple.com

1 0