June 2006 - Darwin-dev - lists.apple.com

Re: Re2: GNU-Darwin: missing code in Darwin2
by proclus＠gnu-darwin.org 05 Jun '06

05 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com While I'm helping with the process of asking for drivers to be released, I would also suggest that the source code for the AppleRTL8169Ethernet gigabit ethernet driver also be released under the terms of a free software agreement such as APSL (especially x86). In this way Darwin-only, open-only developers could have cheap access to gigabit-level transfer speeds, which are are currently not available to them. This would be very helpful for OS testing, backup, and software deployment among other things. Moreover it would be a boon to the usership in general as an excellent example of releasing a Darwin version of a driver which is already free and open source on many other platforms. Perhaps another FOSS Darwin driver already exists which would be suitable to these stated needs. Alternative suggestions are welcome, but they should be free and open source, obviously. Regards, proclus http://www.gnu-darwin.org/ -- Visit proclus realm! http://proclus.tripod.com/ -----BEGIN GEEK CODE BLOCK----- Version: 3.1 GMU/S d+@ s: a+ C++++ UBULI++++$ P+ L+++(++++) E--- W++ N- !o K- w--- !O M++@ V-- PS+++ PE Y+ PGP-- t+++(+) 5+++ X+ R tv-(--)@ b !DI D- G e++++ h--- r+++ y++++ ------END GEEK CODE BLOCK------ _______________________________________________ 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 pgpfdjSHhBPAy.pgp

1 0

Re: StartupItems
by seebs＠plethora.net 05 Jun '06

05 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <r02010500-1046-7F8F2912F4CD11DA99380016CB8A5B30(a)[192.168.1.6]>, fle tcher sandbeck writes: >For example, if the service can only be used when a user is logged >in then there's no point in it launching until a user actually >tries to use it. At that point they are guaranteed to be logged >in. The dependency is built into the fact that the service doesn't >launch until actually needed. Launchd may be having a problem >launching the service at startup, but a better question may be why >is the service configured to launch at startup at all? Sorry, my description was unclear. Conceptually, there is do dependency on a user being logged in. In fact, on a slower machine, the job runs fine either way. On THIS machine, a race condition causes the job to be launched before NetInfo is available, so the UserName key can't be processed, because there's no such user -- indeed, there's no users at all. The effect of automatic login is to change the speed at which various other events happen such that the race condition isn't hit. -s _______________________________________________ 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: StartupItems
by fletcher sandbeck 05 Jun '06

05 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 6/5/06 at 3:37 PM by sgs-lists(a)codewell.com (Steve Sisak): >At 2:15 PM -0500 6/5/06, Peter Seebach wrote: >>launchd "retry until the service shows up, maybe" solution isn't. > >Here's the key. > >Them model is _not_ "retry until the service shows up". > >It's "launch on demand and block until it's available (or fails)". > >The case where that doesn't work is a bug. For example, if the service can only be used when a user is logged in then there's no point in it launching until a user actually tries to use it. At that point they are guaranteed to be logged in. The dependency is built into the fact that the service doesn't launch until actually needed. Launchd may be having a problem launching the service at startup, but a better question may be why is the service configured to launch at startup at all? [fletcher] _______________________________________________ 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: StartupItems
by Steve Sisak 05 Jun '06

05 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Here's the key. Them model is _not_ "retry until the service shows up". It's "launch on demand and block until it's available (or fails)". The case where that doesn't work is a bug. -Steve _______________________________________________ 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… At 2:15 PM -0500 6/5/06, Peter Seebach wrote: launchd "retry until the service shows up, maybe" solution isn't. This email sent to site_archiver(a)lists.apple.com

1 0

Re: StartupItems
by seebs＠plethora.net 05 Jun '06

05 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <149363C0-7D4A-4F20-8E03-D7C7F3432DBC(a)mac.com>, johnsmailinglists@ma c.com writes: >It seems that this is where your desire comes from for sequenced >execution during startup. Don't assume. Ask. Hmm. Speaking of "don't assume, ask"... No, that's not where I got it. This was an example I ran into within the last couple of days, and is a new example. >You would have avoided >this whole thread, might have received a more constructive answer, >and would have gotten the simplest one: "this is a bug, we're working >on it, sorry." But this is only one of the many cases I've run into where the dependency-support solution is trivial and the launchd "retry until the service shows up, maybe" solution isn't. That this particular one is a bug of sorts (and it's technically a misfeature) doesn't change the underlying principle. -s _______________________________________________ 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: StartupItems
by johnsmailinglists＠mac.com 05 Jun '06

05 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Well, if the service that listens on 1234 wasn't ready... then your service would block. How does this result in your service seeing "sorry, no such service? Always, JP _______________________________________________ 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… On Jun 4, 2006, at 7:48 PM, Peter Seebach wrote: In message <55267361-1F7D-437E-82C3-BE7141769224(a)mac.com>, johnsmailinglists@ma c.com writes: Except that sometimes, instead of blocking, you get a connection refused. For instance, as a concrete example, a friend's mini will fail to start a certain job unless you have autologin on. If autologin is off, launchd tries to launch the job, which has a UserName key, before NetInfo is properly up, and FAILS. The job gets cancelled because it can't be started, because the user doesn't exist yet. This is launchd failing to ask for a service that its using: netinfo is up and running. This is a bug, not a design flaw. File a bug report. See Dave Zarzycki's recent post about the jettison logic. (I assume it's a race condition.) It seems that this is where your desire comes from for sequenced execution during startup. Don't assume. Ask. You would have avoided this whole thread, might have received a more constructive answer, and would have gotten the simplest one: "this is a bug, we're working on it, sorry." -- A common mistake that people make when trying to design something completely foolproof is to underestimate the ingenuity of complete fools. - Douglas Adams, Mostly Harmless This email sent to site_archiver(a)lists.apple.com

1 0

launchd bugs (Was: Re: StartupItems)
by Dave Zarzycki 05 Jun '06

05 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Jun 3, 2006, at 7:25 PM, Peter Seebach wrote: This doesn't work; launchd hangs forever at a blue screen without starting Finder... Thank you, davez _______________________________________________ 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… If you can distill this down to a test case, a bug report would be greatly appreciated. This email sent to site_archiver(a)lists.apple.com

1 0

launchd's job jettison logic (Was: Re: StartupItems)
by Dave Zarzycki 05 Jun '06

05 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Our internal bug tracking the mis-feature removal is: 4458207 On Jun 4, 2006, at 12:22 PM, Peter Seebach wrote: <?xml version="1.0" encoding="UTF-8"?> <!DOCTYPE plist PUBLIC "-//Apple Computer//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd"> <plist version="1.0"> <dict> <key>Label</key> <string>localhost.script_name</string> <key>UserName</key> <string>kodak</string> <key>ProgramArguments</key> <array> <string>/path/to/program</string> </array> <key>RunAtLoad</key> <true/> </dict> </plist> This launchd item, placed in /Library/LaunchDaemons, never gets run. Why? -s _______________________________________________ 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/zarzycki%40apple.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… We're planning on (but not committing to) removing the jettison logic from launchd in the Leopard timeframe. Hmm. Well, maybe someone can tell me what I'm doing wrong with launchd. Jun 4 12:06:18 localhost launchd: localhost.script_name: getpwnam ("kodak") failed Jun 4 12:06:18 localhost launchd: localhost.script_name: exited with exit code: 1 Jun 4 12:06:18 localhost launchd: localhost.script_name: 9 more failures without living at least 60 seconds will cause job removal Does the user exist? Why, yes. This is not only a user on the system, but the user that gets automatically logged in once the system is booted. So what happens? Launchd tries to launch this before the netinfo stuff is all up, and ends up yanking it from the job list, not because the script or daemon is problematic... But because the UserName key isn't viable yet. It doesn't matter what's in the ProgramArguments, because the program never gets to being spawned, because the UserName doesn't work. Doesn't this mean launchd ought to, hmm. Maybe check for availability of NetInfo before trying to spawn a job, then killing it for lack of resources? If "don't just suicide when your resources aren't available" is such good advice for programs in a launchd world, why doesn't launchd follow it? This email sent to zarzycki(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

mbuf data access
by jy.cases＠mac.com 05 Jun '06

05 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi all, - How can have access to packet data ? - mbuf_len() on a mbuf (marked as MBUF_EXT) in a mbuf chain is a valid length ? - when modifying such packets do i need to compute some checksums ? Any tips, clarifications,..., will be appreciate. * My MTOD macro is: #define MTOD(m, t) ((t)(mbuf_data(m))) Jean-Yves Cases jy.cases(a)mac.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 writing some NKE UDP filter with packet mangling + packet re- injection, when i have a mbuf (in chain) with flags set to MBUF_PKTHDR i've got an access to data using MTOD(tMbuf, caddr_t), at this point every things works fine but when the mbuf in the chain is marked as MBUF_EXT nothing work at all. This email sent to site_archiver(a)lists.apple.com

1 0

Re: What exactly does DKIOCCDREAD return for kCDSectorAreaSubChannelQ?
by Dan Markarian 05 Jun '06

05 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi Stephen, My source for this MMC-2, section 4.2.2. Dan Le 4 Jun 2006 à 1:53 AM, Stephen F. Booth a écrit : - 4 control bits - 4 address bits - 72 data bits - 16 bits of CRC My source for this MMC-2, section 4.2.2. Thanks, _______________________________________________ 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… Here is my confusion: 96 bits of subcode data = 12 bytes of data. Yet, according to IOCDTypes.h the Q subcode data returned by DKIOCCDREAD is 16 bytes. Can anyone shed some light on this matter? See tables 223 and 224 under READ CD. Its physical structure is a different affair. I am a bit confused as to the exact format of the data returned by the result of a DKIOCCDREAD ioctl() when reading the Q sub-channel of a CD-DA disc. Everything I have read up to this point states that the subcode data from an audio CD is 96 bits long. With this in mind, I expect to see (for a single CD-DA frame): = 96 bits for the general structure, with the data bits looking like (for the program area): - 8 bits each for TNO, INDEX, MIN, SEC, FRAME, ZERO, AMIN, ASEC, AFRAME Here is my confusion: 96 bits of subcode data = 12 bytes of data. Yet, according to IOCDTypes.h the Q subcode data returned by DKIOCCDREAD is 16 bytes. Can anyone shed some light on this matter? I've probably missed something obvious, but I'm stumped. Stephen _______________________________________________ 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/markarian% 40apple.com This email sent to markarian(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0