November 2007 - Darwin-kernel

Re: Writing from the kernel
by Quinn 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Relations, Developer Technical Support, Core OS/Hardware [1] The bug tracking this is <rdar://problem/3524590>. [2] Alas, there are fairly good technical reasons for this. _______________________________________________ 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.… At 9:54 -0600 20/11/07, Kynan Shook wrote: Now, is there any documentation on these interfaces? A google on vnode_open returns very few results... Not really )-: [1] And given the BSD habit of not including formal parameter names in prototypes [2], your only real option is to look at the Darwin source. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Writing from the kernel
by Kynan Shook 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thanks for the gun, I promise I won't shoot myself in the foot. ;-) On Nov 20, 2007, at 5:05 AM, Quinn wrote: In that case I would just write it to a file (-: _______________________________________________ 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.… Now, is there any documentation on these interfaces? A google on vnode_open returns very few results... Kynan It's not that you /can't/ read or write files from the kernel, it's that doing this exposes you to all sorts of crazy deadlock situations. In this case, you're sitting at the block driver level and you end up calling all the way back up to the VFS level. It's very easy to deadlock due to reentrancy. It's also easy to deadlock due to lack of resources. OTOH, for a research project you should be able to make it work. You can get the vnode for a path using vnode_lookup. You can open it using vnode_open. And you can write to it using VNOP_WRITE. You don't even have to take a dependency on the kernel; these are all KPIs. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Writing from the kernel
by Michael Smith 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Nov 19, 2007, at 11:30 PM, Kynan Shook wrote: = Mike _______________________________________________ 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.… For a small research project, I'm interested in all the data written to (and possibly read from) a couple files. From lots of googling and a little playing around, the SampleFilterScheme example can do what I need; I was able to modify it slightly and see the data going through it in system.log. Are you interested in "all the data written to/read from", or "all the data moved to/from the disk"? The two are *very* different things. If the latter, then fine, but if the former then a filter scheme is totally the wrong thing to be using. You want to use dyld interposing and catch the system calls as they are made by the application(s) actually operating on the file(s) in question. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Can not copy files >=4GB. Leopard (9B18)
by Quinn 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Relations, Developer Technical Support, Core OS/Hardware _______________________________________________ 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.… At 15:01 +0400 20/9/07, Alexander Obuschenko wrote: There is a problem when copying a file greater than 4 GB to partitions mounted by our fs-plugin (NTFS for Mac OS X driver) using the Finder application. Nice timing! I debugged this yesterday as part of a DTS incident. Here's the text I sent that developer: After a bit of digging I've discovered that this error indicates that the file is too large for the destination volume. Specifically, the file is 'large' (4 GB or more) but the destination volume does not indicate that it supports large files. Clearly the Finder could be better about reporting this error. I've filed a bug <rdar://problem/5608271> to request a decent error message. To determine if a volume supports large files the Finder checks the bSupports2TBFiles flag returned by PBHGetVolParms. For your volume this flag is not set. File Manager sets the bSupports2TBFiles flag based on the value your file system returns for the VOL_CAP_FMT_2TB_FILESIZE bit in the VOL_CAPABILITIES_FORMAT entries of the ATTR_VOL_CAPABILITIES attribute returned by <x-man-page://2/getattrlist>. The kernel's getattrlist compatibility code sets this attribute based on the value you put in the f_capabilities attribute returned by your VFS plug-in's getattr VFS entry point. In your case you're stating that your volume understands the VOL_CAP_FMT_2TB_FILESIZE capability (the value is set in the "valid" field) but it does not support that capability (the value is clear in the "capabilities" field). If your VFS plug-in does actually support files of 4 GB or larger on disk, I recommend that you claim to support this by returning the VOL_CAP_FMT_2TB_FILESIZE capability set. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Writing from the kernel
by Quinn 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com In that case I would just write it to a file (-: S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Relations, Developer Technical Support, Core OS/Hardware _______________________________________________ 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.… At 1:30 -0600 20/11/07, Kynan Shook wrote: So, what is a good way to get the data out of the kernel? I'm looking for fast and simple here - it's only a project, and will only ever be used on my machine, so minimizing the time spent on implementation is the priority. It's not that you /can't/ read or write files from the kernel, it's that doing this exposes you to all sorts of crazy deadlock situations. In this case, you're sitting at the block driver level and you end up calling all the way back up to the VFS level. It's very easy to deadlock due to reentrancy. It's also easy to deadlock due to lack of resources. OTOH, for a research project you should be able to make it work. You can get the vnode for a path using vnode_lookup. You can open it using vnode_open. And you can write to it using VNOP_WRITE. You don't even have to take a dependency on the kernel; these are all KPIs. You'd better make sure that the file you're writing to is on a different file system that the one that you're filter is running under (-: Keep in mind that the only reason I suggest this approach is that you're working on a research project. For production code I'd recommend a completely different approach. To come up with a good approach I'd have to ask you a bunch of questions about your constraints (like, for example, whether you care about dropping data, and the sort of performance characteristics you're looking for). However, for a research project you might as well take the shortcut. This email sent to site_archiver(a)lists.apple.com

1 0

Writing from the kernel
by Kynan Shook 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thanks much, Kynan [0] http://lists.apple.com/archives/darwin-kernel/2005/Jul/msg00005.html _______________________________________________ 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.… For a small research project, I'm interested in all the data written to (and possibly read from) a couple files. From lots of googling and a little playing around, the SampleFilterScheme example can do what I need; I was able to modify it slightly and see the data going through it in system.log. Now, I'd like to put the data somewhere besides system.log. From the archives, I gather (unsurprisingly) that reading from kernel space is a bad idea[0]. I'm going to assume that if reading is bad, so is writing. And that the standard C libraries (and hence fopen) are excluded from the build so that I don't go and do something stupid. So, what is a good way to get the data out of the kernel? I'm looking for fast and simple here - it's only a project, and will only ever be used on my machine, so minimizing the time spent on implementation is the priority. This email sent to site_archiver(a)lists.apple.com

1 0

Re: missing IOLog entries in /var/log/system.log
by Herb Petschauer 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; bh=Q1K+2FlV0FCO3aoWg2/2Yj9p8rnFiZ9SfdjRhal5THU=; b=boL3dZQwaj7OsXnyOCFLGP15s0V+NfYKQPY4RAGyIDr21R1TSZwtuYeMFLFX5hSXOM668Z8CCdkyH7rRz/ELvLzpB5i03Fb7fDK1UOovYqc+3e1aLAAzyWEtEUroHHtcAGqzR/+6CEx12VBU+Wr8heXksRG7xXfKdVdqVoNJsOU= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=bILP6tsFYjM27Vv8s0bnWV3bahz+SKfGqRy/xvJVoD81qjbbRWG7sTW1C3NX40ydnB4osV7G9ZQTOTgIkcAvLqfCr2lG1T9di9LJ5EPRkbRKbe9i6+1VgXS5zErRB5ucKb8TjFVRsCHZxkrMhnnD+s6QZLmdwe6O/Bd0o7o9va4= None? Never. Is your driver loading during boot? Those messages don't always make it to the system.log. Have you tried booting in verbose mode? You should see them then. Have you tried the FireWireKPrintf tool (from the FireWire SDK)? Cheers, -H. On 19/11/2007, Chris Neustrup <chris(a)asicdesigners.com> wrote: > In developing my IOKit driver, I've relied heavily on IOLog messages in > the system.log. Now there are no longer any entries being placed in the > log. What has happened that I can change to allow messages to be > appended to system.log? > > Thanks in advance, Chris > _______________________________________________ > 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/bogvardi%40gmail.com > > This email sent to bogvardi(a)gmail.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.… This email sent to site_archiver(a)lists.apple.com

1 0

missing IOLog entries in /var/log/system.log
by Chris Neustrup 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com User-agent: Thunderbird 1.5.0.7 (X11/20070925) Thanks in advance, Chris _______________________________________________ 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.… In developing my IOKit driver, I've relied heavily on IOLog messages in the system.log. Now there are no longer any entries being placed in the log. What has happened that I can change to allow messages to be appended to system.log? This email sent to site_archiver(a)lists.apple.com

1 0

Re: external USB drive chown
by mm w 19 Nov '07

19 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; bh=eFxlgBwe09Yq8m+glXwLhXpmFnUza85PUuFnl/t0AJg=; b=MlBvj5w1ezZIJJwp/pGtjo9+lq8zgxBkytXV1AU2doAl1s+xqbmgIR2U5E6qjVnGbK/Q1oXoAwoBNCVEjgbGIjgBFPJSpZTeB/2lSRhFwO/VdN8kvNx2Xl7bGUpPipRl/Q/jjP39oks42F+0ydb2GyAWhFDcQvg8E3yT1SLvWHs= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=TmG6Ee7L+2mzq7eScTOT0OwCM+PszrBWfi2r72T0OFiXUjGMZyfIZ97RUQGsCgHX75mWY0KbQh2Oia8YtGaTnRPFAVEDj+BUnKMXU44TtlC4u3Na56LFIzVSM/OSaDQjwI6oOrbUJPm59bKMpxLEOYvTV3EvNAavzI0exGHGKl8= Hi all, in fact, it does, some people (Internship) opened UI sessions on this devel machine to play video games, the problem is fixed, they left, I have to present all my apologies about this noise. -mmw On Nov 19, 2007 10:01 AM, Michael Smith <drivers(a)mu.org> wrote: > The 'ignore ownership' information is a property of the system, not of > the drive. You need to do this once for every OS installation you > attach the drive to. > > = Mike > > On Nov 18, 2007, at 9:14 AM, mm w wrote: > > > There are no flies on me, anyway, (I should mention "no ``guiguis` > > answer please") > > after a test case, i 've unmount ed and remounted the Volume and the > > problem was fixed, > > this morning, everything was ok (during 2 hours) since the problem > > came again, > > then i 've unmounted and remounted ... GOTO previous ... > > > > *ps: please no ``guiguis` admitted > > > > -mmw > > > > On Nov 18, 2007 8:52 AM, mm w <openspecies(a)gmail.com> wrote: > >> this is the case ... > >> > >> thank you guigui > >> > >> > >> On Nov 17, 2007 11:33 PM, Ethan Clark <eclark(a)monkeyvoodoo.net> > >> wrote: > >>> In Finder, select the drive and do File > Get Info. Uncheck "Ignore > >>> Ownership on This Volume" (it's at the bottom). > >>> > -- echo zapydapntpd.rxltw@nzx | tr a-z@. p-za-o.@ _______________________________________________ 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: external USB drive chown
by Michael Smith 19 Nov '07

19 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com = Mike On Nov 18, 2007, at 9:14 AM, mm w wrote: *ps: please no ``guiguis` admitted -mmw this is the case ... thank you guigui In Finder, select the drive and do File > Get Info. Uncheck "Ignore Ownership on This Volume" (it's at the bottom). _______________________________________________ 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.… The 'ignore ownership' information is a property of the system, not of the drive. You need to do this once for every OS installation you attach the drive to. There are no flies on me, anyway, (I should mention "no ``guiguis` answer please") after a test case, i 've unmount ed and remounted the Volume and the problem was fixed, this morning, everything was ok (during 2 hours) since the problem came again, then i 've unmounted and remounted ... GOTO previous ... On Nov 18, 2007 8:52 AM, mm w <openspecies(a)gmail.com> wrote: On Nov 17, 2007 11:33 PM, Ethan Clark <eclark(a)monkeyvoodoo.net> wrote: This email sent to site_archiver(a)lists.apple.com

1 0