December 2004 - Darwin-dev - lists.apple.com

Re: pthread impl detection
by Eric Albert 21 Dec '04

21 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Dec 21, 2004, at 7:51 AM, lupus(a)ximian.com wrote: -Eric _______________________________________________ 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… I used the disas command in gdb, which apparently uses another codepath to resolve symbols and their location. Using x/i works for me, too, so mistery solved (though you may want to check with the gdb people if that is the intended behaviour of disas). I think 'disassemble' looks in the binary on disk, but x /i prints what's in memory. Then again, I might be wrong about that. :) This email sent to site_archiver(a)lists.apple.com

1 0

Re: pthread impl detection
by lupus＠ximian.com 21 Dec '04

21 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Mail-followup-to: darwin-dev(a)lists.apple.com User-agent: Mutt/1.5.6+20040907i On 12/20/04 Eric Albert wrote: > On Dec 20, 2004, at 7:26 AM, Paolo Molaro wrote: > the best implementation for the specific hardware in use. Short of > copying out that code at app launch and using it, you're likely to end Yes, we jit the code, so we inline the currently used code: unless the implementation changes during runtime (which I guess could be possible, but very unlikely:-) it should be safe. > up with a worse implementation for various systems than the current > code. And while copying out that code at app launch may seem tempting, > even that is dangerous if the location of __pthread_getspecific changes > in the future. There is no dependency on the current location of __pthread_getspecific (pthread_getspecific is disassembled to get its address, so we don't depend on the name either). > >Dump of assembler code for function __pthread_getspecific: > >0xffff82c0 <__pthread_getspecific+0>: nop > >0xffff82c4 <__pthread_getspecific+4>: .long 0x0 > >0xffff82c8 <__pthread_getspecific+8>: .long 0x0 > >0xffff82cc <__pthread_getspecific+12>: .long 0x0 > >[...] > > I'm not sure how you got that output, but here's what I get when I > debug Safari on my PowerBook running 10.3.6: [...] > 0xffff82c0 in __pthread_getspecific () > (gdb) x /6i $pc > 0xffff82c0 <__pthread_getspecific>: rlwinm r5,r3,2,0,29 I used the disas command in gdb, which apparently uses another codepath to resolve symbols and their location. Using x/i works for me, too, so mistery solved (though you may want to check with the gdb people if that is the intended behaviour of disas). Thanks. lupus -- ----------------------------------------------------------------- lupus(a)debian.org debian/rules lupus(a)ximian.com Monkeys do it better _______________________________________________ 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: Darwin rsync modifying source (?)
by Jonathan Paisley 21 Dec '04

21 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 17 Dec 2004, at 13:18, Ben Bond-Lamberty wrote: drwxr-xr-x 20 useradmin AD\group 680 9 Dec 14:51 Users drwxr-xr-x 56 webadmin extension 1904 10 Dec 08:56 Web rsync -a /Volumes/Data /Volumes/backup1/test the Data drive's root becomes: drwxr-xr-x 20 useradmin AD\group 680 9 Dec 14:51 Users drwxr-xr-x 56 labadmin AD\group 1904 10 Dec 08:56 Web I applied the following patch to rsync.c: diff -u -r1.1.1.6 rsync.c --- rsync.c 3 Jan 2004 11:22:00 -0000 1.1.1.6 +++ rsync.c 21 Dec 2004 12:04:56 -0000 @@ -260,7 +260,7 @@ change to groups that the user is a member of */ change_gid = is_in_group(file->gid); } - if (change_uid || change_gid) { + if (!S_ISLNK(st->st_mode) && (change_uid || change_gid)) { if (do_lchown(fname, change_uid?file->uid:st->st_uid, change_gid?file->gid:st->st_gid) != 0) { _______________________________________________ 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… When (and only when) I issue the following command (as root), backing up Data to an attached Firewire drive The only thing I've been able to figure out is that there is a symlink in the User directory tree above that points to the Web folder. If I remove this symlink, the behavior vanishes--rsync doesn't touch the user/group of Web anymore. I have noticed this behaviour too. In my case, I'm using rsyncX (aka rsync_hfs v2.6.0 from the rsyncX CVS). It was causing trouble for me since it was changing a setgid program (slocate) to have the wrong group. Which prevents the attempt to change permissions if the target is a symlink. Darwin follows symlinks with chown, so the permissions on the pointed-to file/dir get changed otherwise. This email sent to site_archiver(a)lists.apple.com

1 0

Re: pthread impl detection
by Eric Albert 21 Dec '04

21 Dec '04

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com On Dec 20, 2004, at 7:26 AM, Paolo Molaro wrote: In order to speedup access to thread local variables in the code generated by the mono JIT, I'd like to inline the code sequence directly instead of calling pthread_getspecific (). Of course I know this is risky, since the implementation could change, Dump of assembler code for function __pthread_getspecific: 0xffff82c0 <__pthread_getspecific+0>: nop 0xffff82c4 <__pthread_getspecific+4>: .long 0x0 0xffff82c8 <__pthread_getspecific+8>: .long 0x0 0xffff82cc <__pthread_getspecific+12>: .long 0x0 [...] -Eric _______________________________________________ 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… You're treading in very dangerous territory here. As you've noticed, the kernel fills out __pthread_getspecific. It does so by including the best implementation for the specific hardware in use. Short of copying out that code at app launch and using it, you're likely to end up with a worse implementation for various systems than the current code. And while copying out that code at app launch may seem tempting, even that is dangerous if the location of __pthread_getspecific changes in the future. Dump of assembler code for function pthread_getspecific: 0x90001de0 <pthread_getspecific+0>: li r4,72 0x90001de4 <pthread_getspecific+4>: ba 0xffff82c0 <__pthread_getspecific> I'm not sure how you got that output, but here's what I get when I debug Safari on my PowerBook running 10.3.6: Breakpoint 1, 0x90001de0 in pthread_getspecific () (gdb) x /2i $pc 0x90001de0 <pthread_getspecific>: li r4,72 0x90001de4 <pthread_getspecific+4>: ba 0xffff82c0 <__pthread_getspecific> (gdb) si 2 0xffff82c0 in __pthread_getspecific () (gdb) x /6i $pc 0xffff82c0 <__pthread_getspecific>: rlwinm r5,r3,2,0,29 0xffff82c4 <__pthread_getspecific+4>: li r0,32754 0xffff82c8 <__pthread_getspecific+8>: sc 0xffff82cc <__pthread_getspecific+12>: add r5,r5,r4 0xffff82d0 <__pthread_getspecific+16>: lwzx r3,r3,r5 0xffff82d4 <__pthread_getspecific+20>: blr This email sent to site_archiver(a)lists.apple.com

1 0

pthread impl detection
by Paolo Molaro 20 Dec '04

20 Dec '04

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com Mail-followup-to: Darwin-dev(a)lists.apple.com User-agent: Mutt/1.5.6+20040907i Hi. In order to speedup access to thread local variables in the code generated by the mono JIT, I'd like to inline the code sequence directly instead of calling pthread_getspecific (). Of course I know this is risky, since the implementation could change, so I wrote some code to detect what pthread_getspecific() actually does: if the code doesn't do what is expected we fallback to the call to pthread_getspecific(). The interesting thing is that when I read the instructions of the pthread_getspecific implementation (ie the __pthread_getspecific code setup by the kernel), they are mostly 0, just the first instruction is a nop. Dump of assembler code for function pthread_getspecific: 0x90001de0 <pthread_getspecific+0>: li r4,72 0x90001de4 <pthread_getspecific+4>: ba 0xffff82c0 <__pthread_getspecific> Dump of assembler code for function __pthread_getspecific: 0xffff82c0 <__pthread_getspecific+0>: nop 0xffff82c4 <__pthread_getspecific+4>: .long 0x0 0xffff82c8 <__pthread_getspecific+8>: .long 0x0 0xffff82cc <__pthread_getspecific+12>: .long 0x0 [...] It's a surprise that the cpu can execute the code even if it appears as nop/zeroes when read. I tried to mprotect the relevant page, but that returns permission denied: it shouldn't be read-protected though, or I'd get a SEGV reading it. Anyway, for informational purpouses it would be interesting to know why reading the memory area returns odd results. Forcing the code to use the spr register 259 (as should be on the G5 machine) works and gives the expected speedup. On the practical side: anyone has suggestions on how to reliably detect which code sequence to use? Depending just on the processor type is more fragile than actually checking what the code does and we'd have to use the slower function call on Mac OSX... Thanks. lupus -- ----------------------------------------------------------------- lupus(a)debian.org debian/rules lupus(a)ximian.com Monkeys do it better _______________________________________________ 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

Licensing and cross-licensing questions
by Stéphane Sudre 19 Dec '04

19 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Namely the Resources: EnglishToISO.strings and Language.strings I have some questions related to this: TIA _______________________________________________ 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… As of today, I'm using in one of my project 2 files from /System/Library/PrivateFrameworks/IntlPreferences.framework. I'm loading these 2 files which are installed by the OS so it's not part of the source code. Now, if, for instance, these 2 files were to vanish in the future (because you never know what could happen with a private Framework), it could be a problem for my code. One of these files (EnglishToISO.strings ) is a textual representation of code available in CFBundle_Resources.c which is part of the Darwin source code (CoreFoundation (Lite) project). The second one is a text file without any patented information or technic AFAIK. 1. Is it possible to add to my project source the file which is a textual representation of some Darwin source code considering that in that case, I would add the APPLE_LICENSE_HEADER_START to it so copyrights and origin are well known? (The APSL license would be added to the list of licenses I'm already providing with the binary). 2. Is adding an APSL licensed file into a BSD licensed project contaminating it? Would it become an APSL project? This is still not clear to me as IANAL. 3. As for the second file, is it possible to distribute it with the source code by adding some specific license? This email sent to site_archiver(a)lists.apple.com

1 0

Darwin rsync modifying source (?)
by Ben Bond-Lamberty 17 Dec '04

17 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Microsoft-Entourage/10.1.4.030702.0 (First, if this is not the corrrect place to post this question, please accept my apologies. Neither the OS X server listserv nor googling has turned up anything.) Are there any known issues with rsync under Darwin and symlinks? I have a "Data" drive, attached to a 10.3.6 OS X Server on a G5 Xserve, that looks like this at its root: drwxr-xr-x 20 useradmin AD\group 680 9 Dec 14:51 Users drwxr-xr-x 56 webadmin extension 1904 10 Dec 08:56 Web When (and only when) I issue the following command (as root), backing up Data to an attached Firewire drive rsync -a /Volumes/Data /Volumes/backup1/test the Data drive's root becomes: drwxr-xr-x 20 useradmin AD\group 680 9 Dec 14:51 Users drwxr-xr-x 56 labadmin AD\group 1904 10 Dec 08:56 Web (The AD\group is a Active Directory group.) In other words, rsync seems to be modifying the user/group! This happens even if nothing has changed on Data--e.g., I issue an rsync command, chown the Web folder back to its correct permissions, and rsync again. The only thing I've been able to figure out is that there is a symlink in the User directory tree above that points to the Web folder. If I remove this symlink, the behavior vanishes--rsync doesn't touch the user/group of Web anymore. Thanks for any insight- Ben _______________________________________________ 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: inet_aton family in kernel space
by Justin Walker 16 Dec '04

16 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Dec 15, 2004, at 20:31, Shantonu Sen wrote: There's 'safe' and there's "safe"; I was referring to the former :-}. Cheers, Justin On Dec 15, 2004, at 8:28 PM, Justin Walker wrote: On Dec 15, 2004, at 18:58, Andrew White wrote: int inet_aton(char * cp, struct in_addr * pin) Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | If you're not confused, | You're not paying attention *--------------------------------------*-------------------------------* _______________________________________________ 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… Uh, if it's not in a public header file in /S/L/F/Kernel.framework, I wouldn't call it "safe to use" by any means... Is there a version of inet_aton which is safe to use inside an IOKit device driver? How do I include it? The kernel appears to export it, so it's safe to use. I believe the implementation uses this signature: and is found in xnu/bsd/kern/netboot.c. This, of course, may change across releases. -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | When LuteFisk is outlawed | Only outlaws will have | LuteFisk *-------------------------------------- *-------------------------------* _______________________________________________ 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/ ssen%40opendarwin.org This email sent to ssen(a)opendarwin.org This email sent to site_archiver(a)lists.apple.com

1 0

Re: inet_aton family in kernel space
by Vincent Lubet 16 Dec '04

16 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Dec 15, 2004, at 6:58 PM, Andrew White wrote: Vincent _______________________________________________ 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… Is there a version of inet_aton which is safe to use inside an IOKit device driver? How do I include it? If you're inside an IOKit device I recommend that you including your own version of inet_aton --there are a lot of open source implementation of inet_aton available. You could link your IOKit module with the inet_aton version xnu/bsd/kern/netboot.c but that would make your IOKit module not compatible with Tiger. This email sent to site_archiver(a)lists.apple.com

1 0

Re: inet_aton family in kernel space
by Shantonu Sen 16 Dec '04

16 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Shantonu On Dec 15, 2004, at 8:28 PM, Justin Walker wrote: On Dec 15, 2004, at 18:58, Andrew White wrote: int inet_aton(char * cp, struct in_addr * pin) Regards, Justin _______________________________________________ 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/ssen%40opendarwin.org _______________________________________________ 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… Uh, if it's not in a public header file in /S/L/F/Kernel.framework, I wouldn't call it "safe to use" by any means... Is there a version of inet_aton which is safe to use inside an IOKit device driver? How do I include it? The kernel appears to export it, so it's safe to use. I believe the implementation uses this signature: and is found in xnu/bsd/kern/netboot.c. This, of course, may change across releases. -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | When LuteFisk is outlawed | Only outlaws will have | LuteFisk *-------------------------------------- *-------------------------------* This email sent to ssen(a)opendarwin.org This email sent to site_archiver(a)lists.apple.com

1 0