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

Re: Apple Instrumentation libraray
by Philip Aker 20 Mar '07

20 Mar '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com ftp://ftp.apple.com/developer/Development_Kits/Instrumentation_SDKhqx ftp://ftp.apple.com/developer/Development_Kits Look in the Obsolete_Unsupported folder Philip Aker philip_aker%40mac.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.… On 2007-03-19, at 21:31:46, MFAWZY wrote: it seems it does not exist there anymore?! Any idea. This email sent to site_archiver(a)lists.apple.com

1 0

Apple Instrumentation libraray
by MFAWZY 20 Mar '07

20 Mar '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thread-index: AcdqqKqj0Wsnm0WlRqiDWkowxofv0Q== _______________________________________________ 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: Executable entry point
by Jonas Maebe 20 Mar '07

20 Mar '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On 19 Mar 2007, at 17:24, Filipe Cabecinhas wrote: My compiler outputs files for nasm so I compile them like this: nasm -fmacho hello.asm ld -o hello hello.o -lrts -L../build -e _start -lc So it links with libc. But, when I try to run it, I get: (gdb) disassemble dyld_stub_write Dump of assembler code for function dyld_stub_write: 0xa0011230 <dyld_stub_write+0>: jmp 0x900101b0 <write> End of assembler dump. (gdb) r Starting program: /Users/filcab/dev/compiladores/tests/hello Jonas _______________________________________________ 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.… Program received signal EXC_BAD_INSTRUCTION, Illegal instruction/ operand. 0x00004005 in dyld_stub_write () (gdb) x /i $eip 0x4005 <dyld_stub_write>: hlt I'm guessing (maybe I'm wrong) that the dynamic linker isn't loading libc. Any clues? This suggests to me that dyld doesn't get initialised properly. Have you tried linking in /usr/lib/crt1.o already? In that case, you may not be able to specify an alternate entry point though. Another option may be to add the source code of crt1.o to your program so you can change its symbol names for entry point and where it jumps to in your program after it finishes initialising, but note that it was quite a convoluted mess the last time I looked at it. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Executable entry point
by Filipe Cabecinhas 19 Mar '07

19 Mar '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:in-reply-to:references:mime-version:content-type:message-id:cc:content-transfer-encoding:from:subject:date:to:x-mailer; b=UVMwZ3NgS6Kh3DLQthSscCGbIHfAfvl3A3iU/b5+TTGF1UkLbj2BU5UIxLjJ3q8EJDRrgBfAL+GP8tvmChCd1QZ0fBESx6HJzKECWiR5318NxDJ+hdGVRZFz81KFke3VW/+EY4DSB9lu1aHFB9gUyf+xthyjy5hi0ZsWGoy4EK8= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:in-reply-to:references:mime-version:content-type:message-id:cc:content-transfer-encoding:from:subject:date:to:x-mailer; b=Ij9fImUvIk2wraDvRkR0lDcbTyE2pTaJVwY+N1GTcaa2gx0fUO05eTE81PVqJjr69+VZKO9N6FTZmguFx+LE+1FM8CdMKKIGLvPvDQCTW/8spG7GYk8mdC7S/S97xWNefvctP7e27wCzfsOo1P6w0XsEtCLZZ4FJ+sTL7JPjn68= My compiler outputs files for nasm so I compile them like this: nasm -fmacho hello.asm ld -o hello hello.o -lrts -L../build -e _start -lc So it links with libc. But, when I try to run it, I get: (gdb) disassemble dyld_stub_write Dump of assembler code for function dyld_stub_write: 0xa0011230 <dyld_stub_write+0>: jmp 0x900101b0 <write> End of assembler dump. (gdb) r Starting program: /Users/filcab/dev/compiladores/tests/hello Any clues? Thanks for the help Filipe Cabecinhas On 18 Mar, 2007, at 12:30, Jonas Maebe 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'm now trying to write the run-time library using libc (When I corrected the sysenter_trap it stopped doing system calls, so I tried this way) but I have another problem... This time with the linker. Program received signal EXC_BAD_INSTRUCTION, Illegal instruction/ operand. 0x00004005 in dyld_stub_write () (gdb) x /i $eip 0x4005 <dyld_stub_write>: hlt I'm guessing (maybe I'm wrong) that the dynamic linker isn't loading libc. In general, it's pretty easy to build your run time library on top of libc rather than on top of system calls (we also did it for Mac OS X, while our Linux version uses system calls because there system calls are pretty much guaranteed to be backwards compatible forever, while glibc breaks backwards binary compatibility every couple of years). This email sent to site_archiver(a)lists.apple.com

1 0

Re: Executable entry point
by Filipe Cabecinhas 18 Mar '07

18 Mar '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:in-reply-to:references:mime-version:content-type:message-id:cc:content-transfer-encoding:from:subject:date:to:x-mailer; b=PlNvt9Fj6IHcJAr+CUlXdeREdflQ4nWgy+3yx96qoDS/eojYWVV/8+zblJ0EdiCvL1hbC7/muOUzaq+r86B4m5g92yorsYqEEz3vecdVelTDOHSnThZQ/qjtSKTGoj/ltCNvWZTNYv5PG5UbDOxJEW1xiWrr74WCtXfClzb/xjw= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:in-reply-to:references:mime-version:content-type:message-id:cc:content-transfer-encoding:from:subject:date:to:x-mailer; b=e//x87Va/bvPiibVJ4LJlC909pCTTqqY8Rv6cXIcGdoVnYj/PsrxMshWVQfMvjNTgbcf9LscyHsUNJLIEEy5N5l1ml59yqPaDX0mpoMimHr65l3J2V3jGQzKCP2OYUSmxEH8+/HdqLisMFLXofSTAdjyIN53BAe1wbDh1xR1Fls= On 18 Mar, 2007, at 12:20, Geoffrey Lee wrote: No, you missed something: sysenter_trap looks like this: __sysenter_trap: 90110ac0 popl %edx 90110ac1 movl %esp,%ecx 90110ac3 sysenter So, %ecx contains the saved stack pointer and %edx contains the return address. Thanks for the help Filipe Cabecinhas _______________________________________________ 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.… Yes, I missed that. I think I removed it when I removed the stuff for PIC. I don't know why... By the way, I am not quite sure if doing things directly is such a good idea ... can you not go through libc? - gl Yes, doing things directly is a Bad Thing (TM). But the thing is... Our compiler class has to use the library they give us, and it is made for linux (with the int 0x80 stuff). So, in Mac OS X, I can't link to libc so I have the setup as similar as I can. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Executable entry point
by Jonas Maebe 18 Mar '07

18 Mar '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On 18 Mar 2007, at 12:02, Filipe Cabecinhas wrote: Jonas _______________________________________________ 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.… As far as I could see (by going through the source of Darwin, when I want to call a system call directly (I can't link the output of my compiler with the libc, I have to link it with my run-time library) I should do the following: Note that in this case the programs you create are only guaranteed to run on the exact system revision you are compiling on, because the syscall interface on Mac OS X is undefined and can change at will (unlike libc, which is defined). In general, it's pretty easy to build your run time library on top of libc rather than on top of system calls (we also did it for Mac OS X, while our Linux version uses system calls because there system calls are pretty much guaranteed to be backwards compatible forever, while glibc breaks backwards binary compatibility every couple of years). This email sent to site_archiver(a)lists.apple.com

1 0

Re: Executable entry point
by Geoffrey Lee 18 Mar '07

18 Mar '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com User-agent: Mutt/1.5.12-2006-07-14 On Sun, Mar 18, 2007 at 11:02:55AM +0000, Filipe Cabecinhas wrote: > Thank you very much, now it runs :D > > I have now another problem. > > As far as I could see (by going through the source of Darwin, when I > want to call a system call directly (I can't link the output of my > compiler with the libc, I have to link it with my run-time library) I > should do the following: > > push the arguments to the stack(for example (write): push nbytes, > push str, push fd) > set eax to the system call number (4) > call __sysenter_trap > > sysenter_trap: > put the esp in ecx > sysenter > > I think I should call the trap so the system call has an address to > return to, right? > No, you missed something: sysenter_trap looks like this: __sysenter_trap: 90110ac0 popl %edx 90110ac1 movl %esp,%ecx 90110ac3 sysenter So, %ecx contains the saved stack pointer and %edx contains the return address. By the way, I am not quite sure if doing things directly is such a good idea ... can you not go through libc? - gl > But my problem is, when I return from sysenter, it goes to a weird > position instead of going to the address after the call __sysenter_trap. > > Am I doing anything wrong? > > > Thanks for the help > > Filipe Cabecinhas > > > On 18 Mar, 2007, at 07:50, Geoffrey Lee wrote: > > >> > >>The question is: how does darwin know where is the entry point of the > >>binary and how can I choose it? > >> > > > >Use -e flag in ld(1). It is actually stored in the thread state, > >the kernel loads this thread state when the binary is loaded, you can > >see for yourself with otool -l, srr0 on ppc and eip on 32bit x86. > > > > > > - gl > > _______________________________________________ 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: Executable entry point
by Filipe Cabecinhas 18 Mar '07

18 Mar '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:in-reply-to:references:mime-version:content-type:message-id:cc:content-transfer-encoding:from:subject:date:to:x-mailer; b=jw+bBWkHOSKvI9zGwLGHIH6vudR2vIeeem6q1JfoblT3cOZn8I8WXI5DVWcCUGQTa0cz+0XoAsHZE2smf5r0EEVO3vZfgpjLuIon0DBEmXzTo45tqqQrP+wmnZDmuzDfnO6I6xB7LZ/4a14oBlIj3v6AbfoR3SG3KhjBx+4WSEs= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:in-reply-to:references:mime-version:content-type:message-id:cc:content-transfer-encoding:from:subject:date:to:x-mailer; b=fDRUO/DI2Czn+wMIS5VyZiXtwMnzSikVwv9tJPqUIsORRh3mTMzHxNx4BKw210I8clF/mJAp1PfhhO7njF+oeyCzttw9+qQnvHM2eB/AY3jIytFQEs1TFDodrw7SSGB8cqR4r2q6csqb6o5CNJxPzZrnDqx3Gef3wQdpuFNtobY= Thank you very much, now it runs :D I have now another problem. sysenter_trap: put the esp in ecx sysenter Am I doing anything wrong? Thanks for the help Filipe Cabecinhas On 18 Mar, 2007, at 07:50, Geoffrey Lee wrote: The question is: how does darwin know where is the entry point of the binary and how can I choose it? Use -e flag in ld(1). It is actually stored in the thread state, the kernel loads this thread state when the binary is loaded, you can see for yourself with otool -l, srr0 on ppc and eip on 32bit x86. _______________________________________________ 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.… As far as I could see (by going through the source of Darwin, when I want to call a system call directly (I can't link the output of my compiler with the libc, I have to link it with my run-time library) I should do the following: push the arguments to the stack(for example (write): push nbytes, push str, push fd) set eax to the system call number (4) call __sysenter_trap I think I should call the trap so the system call has an address to return to, right? But my problem is, when I return from sysenter, it goes to a weird position instead of going to the address after the call __sysenter_trap. - gl This email sent to site_archiver(a)lists.apple.com

1 0

Re: Executable entry point
by Geoffrey Lee 18 Mar '07

18 Mar '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com User-agent: Mutt/1.5.12-2006-07-14 On Sun, Mar 18, 2007 at 01:43:48AM +0000, Filipe Cabecinhas wrote: > Hi > > I'm trying to port a compiler we had in class for Mac OS X but I'm > having a little problem with the executable's entry point. > > Our compiler translates from the source language to assembly, then I > would assemble it with NASM, link it with our run-time library and > then I could run it. > > I tried to port the run-time library (changing the "int 0x80" to > "sysenter"s and some other house-keeping stuff) and I'm trying to > test it to see if that went well. > > The problem is: After the assembling and linking part (both go well, > without any kind of warnings or errors), I try to execute it but it > doesn't run. I was thinking darwin would call the _start function but > maybe that's not the case. I also tried googling to see if I could > tell nasm/ld which function was the entry point but I had no luck. > > The question is: how does darwin know where is the entry point of the > binary and how can I choose it? > Use -e flag in ld(1). It is actually stored in the thread state, the kernel loads this thread state when the binary is loaded, you can see for yourself with otool -l, srr0 on ppc and eip on 32bit x86. - gl > If I should post in another list please tell me which. By reading the > titles of the lists, this one seemed the best. > > Thanks for the help > > Filipe Cabecinhas > > _______________________________________________ > 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/glee%40cse.unsw.edu.au > > This email sent to glee(a)cse.unsw.edu.au > _______________________________________________ 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

Executable entry point
by Filipe Cabecinhas 18 Mar '07

18 Mar '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:mime-version:content-transfer-encoding:message-id:content-type:to:from:subject:date:x-mailer; b=FxUNiwCqfDK6WOcErmKWMAMW7me+VxEzMBtL33enpNUlcOndUmrrkN+rVCy+HBcN2d+wvqjSGeKsOkRgkLhGYlaqkahxMf89f5O5jsJdJIO0LBmMAdOcyeHnPgs7hz+5qa2mNamgGwDdBnkDIL4L/NGkPBa1wHcxilqO91Rt5g4= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:mime-version:content-transfer-encoding:message-id:content-type:to:from:subject:date:x-mailer; b=ouqRgtPLAIcgCB3lwiObV2Im5d8dvbx1rKUGdAGIW4x7a/8je3ihoOxPkSOJ9Cg6BvyLo4GU3Z+Taez5lHtvRZLDsP3H31s9VuEi3C6QJ/cTxCO/z4pJUVI2ZcK+TLrMro/EBGF/VpV3VAhRn8CuY9cboC0ueIwaGSxYLmPEcrk= Hi Thanks for the help Filipe Cabecinhas _______________________________________________ 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'm trying to port a compiler we had in class for Mac OS X but I'm having a little problem with the executable's entry point. Our compiler translates from the source language to assembly, then I would assemble it with NASM, link it with our run-time library and then I could run it. I tried to port the run-time library (changing the "int 0x80" to "sysenter"s and some other house-keeping stuff) and I'm trying to test it to see if that went well. The problem is: After the assembling and linking part (both go well, without any kind of warnings or errors), I try to execute it but it doesn't run. I was thinking darwin would call the _start function but maybe that's not the case. I also tried googling to see if I could tell nasm/ld which function was the entry point but I had no luck. The question is: how does darwin know where is the entry point of the binary and how can I choose it? If I should post in another list please tell me which. By reading the titles of the lists, this one seemed the best. This email sent to site_archiver(a)lists.apple.com

1 0