October 2009 - Darwin-dev - lists.apple.com

Re: launchd bug?: System: job_mig_spawn() can't find its session!
by Damien Sorresso 31 Oct '09

31 Oct '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Oct 31, 2009, at 9:02 AM, lbland wrote: > hi guys!- > > I get this in my console: > > Oct 31 11:54:31 h003 com.apple.launchd[1]: System: Bug: launchd_core_logic.c:8686 (23932):0 > Oct 31 11:54:31 h003 com.apple.launchd[1]: System: job_mig_spawn() can't find its session! > Oct 31 11:54:31 h003 my_process[225]: spawn_via_launchd() failed, errno=5 label=[0x0-0x75f75f].com.company.another path=a_path flags=1 > > when using the launchd services API in a program. You're talking about LaunchServices, I believe. Those APIs are not safe to call from a daemon. You should put this code in a launchd agent instead. > It only seems to happen when I logout and then login to a different account. It did not seem to happen under Mac OS X 10.6.0, but is happening under 10.6.1. > > Maybe 10.6.2 will fix it?? > > Google seems to indicate that it is not too uncommon (not unique) of a problem. This is correct behavior. > I know I've been a bit vague here, but feel free to reply with vague hints on what is wrong :-) Daemons cannot manipulate the GUI, which is what the LaunchServices APIs do (by launching an application). Application launching only makes sense in the context of a user session, but you're calling into the API from the system context. Think of it this way. When you're writing a daemon, you must assume that the machine is headless (no user logged in). If you have access to the WWDC session videos, this topic was covered in greater detail in the "Designing for launchd" session. -- Damien Sorresso BSD Engineering Apple Inc. _______________________________________________ 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

launchd bug?: System: job_mig_spawn() can't find its session!
by lbland 31 Oct '09

31 Oct '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com hi guys!- I get this in my console: Maybe 10.6.2 will fix it?? thanks!- -lance _______________________________________________ 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… Oct 31 11:54:31 h003 com.apple.launchd[1]: System: Bug: launchd_core_logic.c:8686 (23932):0 Oct 31 11:54:31 h003 com.apple.launchd[1]: System: job_mig_spawn() can't find its session! Oct 31 11:54:31 h003 my_process[225]: spawn_via_launchd() failed, errno=5 label=[0x0-0x75f75f].com.company.another path=a_path flags=1 when using the launchd services API in a program. It only seems to happen when I logout and then login to a different account. It did not seem to happen under Mac OS X 10.6.0, but is happening under 10.6.1. Google seems to indicate that it is not too uncommon (not unique) of a problem. I know I've been a bit vague here, but feel free to reply with vague hints on what is wrong :-) This email sent to site_archiver(a)lists.apple.com

1 0

not seeing dynamic library page-ins
by Joel Reymont 30 Oct '09

30 Oct '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:from:content-type :content-transfer-encoding:subject:date:message-id:to:mime-version :x-mailer; bh=iKegOFjLosnTR8pf+5aod8M43RYM+9lQSug8dqE7k4g=; b=NjjxBsMh/h59WecH9yoxWRBFgnM3wTcqu3znFKsoIu10dKYlXnc2dvlU8sX9ZA4D7Q 6nQmUw7rjt1WKsWoeCJ3WLEHe0AFkDUZpcmMNYB6qNzZJrULIDeUbz6Cx83iY2HPnBad 3uIPFEaOx6sL/vebK56hfj/JO6rQm1MW+fH9I= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=from:content-type:content-transfer-encoding:subject:date:message-id :to:mime-version:x-mailer; b=TaUIoBdJXSEdVgDzkfS86xc5FnW8VYkgPp44VXUidM/wVQ8l+jZVCkxvQdGOGNaOn/ puoyZzlsnWJyasI0RpwLeBn7G8Af8oRHVvVGLCNZzup3+zykfmE2MqO51lsRrddw3tAU 6xXyaotqtX2Rk0CBMTT7HdpzfUxYvl86fPEYE= I'm tracking page-ins in dynamic libraries used by Firefox. I'm on Snow Leopard and using these DTrace probes: vminfo::vnode_pagein:pgpgin { printf("page-in\n"); } fbt::vnode_pagein:entry { self->v_name = stringof(((vnode_t)arg0)->v_name); printf("vnode name: %s\n", self->v_name); } I see the library being opened by the dynamic linker and Firefox comes up just fine. Trouble is, I don't see any page-ins. I thought this may be due to the Unified Buffer Cache but disabling the cache on a given library while keeping the while open has no effect. Where do the code pages come from and what am I missing? Thanks, Joel --- firefox for android! http://wagerlabs.com _______________________________________________ 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: F_GLOBAL_NOCACHE not global?
by Michael Smith 30 Oct '09

30 Oct '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com _______________________________________________ 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

F_GLOBAL_NOCACHE not global?
by Joel Reymont 30 Oct '09

30 Oct '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:from:content-type :content-transfer-encoding:subject:date:message-id:to:mime-version :x-mailer; bh=eBdzu3s8481q0b8NucKbV5DTI3nLbUbjKyv0KFFku2k=; b=wU4JV6Pr6w2aL0mjx5/wyFOM/PancNIiljGh7N6gW3Ulw8EQccxocDCOxUvBM4CkDI 1zgmvUCzbqmspTZqJvNBcCN6xkw4GFGkkuI+unJlC3Zh2eVUz2PO9q74XawH8t4A541F qgaDE0LFU54MnvzOhuw6bd/3CcOMKTwfWlw9M= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=from:content-type:content-transfer-encoding:subject:date:message-id :to:mime-version:x-mailer; b=e1YM/zBrVyT6GevuDyR794eLAK2wPi/Wh/wi6nTcB33c15zLSAQezKLpG8E0ZW5Khu R+y5QQz+HBL0bmAWsqxJU5XmZRBvyaUsPLsI7WqBREQ3487KvP/e/vD0D2kr7M/yniAL AOHCpqoNQbDQaUExkK5EmMSEVz+yrxOKMDKd0= Is F_GLOBAL_NOCACHE really global or am I missing something? Thanks, Joel --- firefox for android! http://wagerlabs.com _______________________________________________ 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'm using fcntl(fd, F_GLOBAL_NOCACHE, 1) to set the global "no cache" flag on a file. The changes do not seem to persist once I exit the process that sets the flag. This email sent to site_archiver(a)lists.apple.com

1 0

incompatible libstdc++.6 and libSystem.B
by Jörg Richter 28 Oct '09

28 Oct '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi, The application is running fine on 10.4 and 10.5, but not on 10.6. It dumps core as soon as an exception is thrown. This is the top of the stack, when it happens: But doing nm on libSystem I get this symbols: 0019f9b8 S $ld$hide$os10.4$__Unwind_GetIP 0019f9b9 S $ld$hide$os10.5$__Unwind_GetIP 0006fb50 T __Unwind_GetIP Are the above options enough to request 10.4 compatibility? Thanks for your help, Jörg _______________________________________________ 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 am developing an C++ application on 10.5. But I use a custom built FSF GCC 4.4.1. Because I want to target >=10.4 I built with this flags: -isysroot /Developer/SDKs/MacOSX10.4u.sdk -Wl,-syslibroot,/ Developer/SDKs/MacOSX10.4u.sdk -mmacosx-version-min=10.4 0 libstdc++.6.dylib 0x01df849b __cxa_throw + 59 1 libSystem.B.dylib 0x984fab70 _Unwind_GetIP + 32 2 libstdc++.6.dylib 0x01df7d01 __gxx_personality_v0 + 287 3 libgcc_s.1.dylib 0x01f10f74 _Unwind_RaiseException + 148 4 libstdc++.6.dylib 0x01df84e0 __cxa_throw + 128 libstdc++ and libgcc_s are from my custom built GCC. libSystem.B from 10.6. As you can see, frame 2 (__gxx_personality_v0) calls into libSystem.B.dylib (_Unwind_GetIP). I think this should not happen because they are not compatible. Only 10.6 has _Unwind_GetIP in libSystem.B.dylib I think the $ld$hide$os10.4$ prefix means, that this function should not be considered when loading a program targeted for 10.4. But it obviously is. This is where I got stuck. How does the dynamic linker know which OS version a program was built for? Can this be queried from the command-line? I *think* otool should do it, but I was unable to find it in its output. This email sent to site_archiver(a)lists.apple.com

1 0

tracking io from dynamic library code page-ins
by Joel Reymont 27 Oct '09

27 Oct '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:from:content-type :content-transfer-encoding:subject:date:message-id:to:mime-version :x-mailer; bh=xyvoeM/SSm60t8nfkOkcIVSNtW4qxZpSAuQjHm6DL5c=; b=lHB3MHwByCnHD5Hlk7AfMDnVCKsy/nGT8wUXYV/FUKow7BTmR7nwCuy/u5gBeAukG9 67cflTee+Ujb0Br32bByCQdwHd5rJMZO5QqOuD2IUw5jxjWQh6eVBAopP3lKJWIqZyKm lrF3l6VBsWd8FFeFzkFCWv9hgM2JPRqonsAV0= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=from:content-type:content-transfer-encoding:subject:date:message-id :to:mime-version:x-mailer; b=A3j8ABq6Ov+77dx4xjCgtYtQ1snHUr7n24f+ZD8uT1ZARv1t1yUZsQbkby9mtErg7Z /mh29tW/82YYAjfDhi50ltMbdGBF9L44hhqw63Wy+yeS3U8txp5o7x4Jf6CkMxXOJ8ZY iY8+C3whegrCU53CERmmgwHiPd3DGrptvY2PY= I was asked what I'm trying to achieve so here's a summary... I'm trying to rearrange the symbols in the dynamic libraries that Firefox uses. The libraries are mmap-ed but the pages are hit in an order that results in bad IO performance. We know that we are taking a hit from the dynamic library loads but we don't know what the access patterns are and that's what I'm trying to find out. Once I know in what order the code pages are loaded from disk by the virtual memory manager, I can try to translate that to symbols and then rearrange those symbols such that the code pages are loaded from disk more or less sequentially. The easiest approach seems to be to piggyback on vnode_pagein from bsd/vm/vnode_pager.c but I can't find a way to associate the vnodes with the path names that were originally passed to open. Any suggestions? --- firefox for android! http://wagerlabs.com _______________________________________________ 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

vnode from fd
by Joel Reymont 27 Oct '09

27 Oct '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:from:content-type :content-transfer-encoding:subject:date:message-id:to:mime-version :x-mailer; bh=YLSeIKok0g2lHbEYF5K5xPcqvUeXHU5h7yDxapdxB8o=; b=TLL7uzOq8ioJV3HR/pAehJIwcepEd4a7KEQuxfqzAFA6Z7XOXJSYp/spDbtHosm/Ve K4TDb3q6HGhIEsD8U1jtJIySnu4gBd8kk7UNZkGabceL6DJNAyuXwOM16gjAFiRB/9IL SsqeAWyTZpkue5GeAfbV4ZEZWc8kkIdlTHM/I= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=from:content-type:content-transfer-encoding:subject:date:message-id :to:mime-version:x-mailer; b=AhC3hMEJiInDDU4OBvsyTs27Rb9sd7jdLO21AUm7gaEt3wJ+0PkVb2zAiEIP7Nqhfr M9nHLZk4C+QsOp7uZ+5cs7TPmqOwZFde8M8OPiToTDsS3hpybPwtGgpDKvpWzD2FjpO5 XJo1DsUaUhCVK4gb0BMb70GDmgM4MhAJlJ34c= I'm trying to capture the vnode of a file in dtrace. Any suggestions? Thanks, Joel --- firefox for android! http://wagerlabs.com _______________________________________________ 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 couldn't find a probe in fbt that would give me access to the path of the file and the vnode so I thought that perhaps I could grab the file via the file descriptor. Is this possible at all? Basically, I need to map file names to vnodes and I need to do it in a probe within dtrace. The reason for this is to be able to track the vnode in fbt probes like vnode_pagein. This email sent to site_archiver(a)lists.apple.com

1 0

Re: TRIM in the drivers and OS for SSD?
by lbland 27 Oct '09

27 Oct '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Chris- On Oct 27, 2009, at 12:38 PM, Chris Sarcone wrote: p.s. - wish I could help, but I already have too much on my plate :-) thanks!- -lance _______________________________________________ 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… We're currently investigating TRIM support. Thank you. That is all I wanted to know. Now I can stop standing by the water cooler and get back to work. :-) This email sent to site_archiver(a)lists.apple.com

1 0

Re: TRIM in the drivers and OS for SSD?
by Eli Bach 27 Oct '09

27 Oct '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:subject:mime-version :content-type:from:in-reply-to:date:content-transfer-encoding :message-id:references:to:x-mailer; bh=xJuRqlB3NWv+oWDAZDXQ9ZaXCQfGJ+Rr3enyVM1FEzs=; b=gpP2BakkUgtwPpnFSTvfy38c5cTW9aYZGOR6PyQIjurZxyAEjbS4Ni/t8imTntLorX O+hH/SBE4X82UpbDkXywWTggdR7SrchZe72+d+xgj0J8ZwTUMn4aPJm2Wh4b1Nc8vXLs xF9xOSFjBI2Y2IGQoeUYNA7WTxwxTnyRS2dUg= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=subject:mime-version:content-type:from:in-reply-to:date :content-transfer-encoding:message-id:references:to:x-mailer; b=r8U4Nhz9ZyNeCucHf1yqG5VQa/XfkdsF6VXR3vNR2A7Ahq3d1wTxDQdqjn1ecOTE5r f6bk3k02Qh6ykbOZYP+ZLznWTVnGEtO3ZOBtsQuw+ipazpb0t3+zjNZPoA2iekUyzY9N H/9oBSx+zfCmkZQgjfDCD9hLTQkUC+4v+iTZ0= On Oct 27, 2009, at 10:38 AM, Chris Sarcone wrote: Hi Lance -- Fired! No giving us false hope over what might be coming... :-0 Eli _______________________________________________ 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… Now that Windows 7 has TRIM, SSDs are finally looking very attractive. Any info. on TRIM in darwin/Mac OS X drivers? We're currently investigating TRIM support. This email sent to site_archiver(a)lists.apple.com

1 0