April 2008 - Darwin-kernel - lists.apple.com

Re: ioctl(_IOR,'t',19,44) messages
by Michael Crawford 02 Apr '08

02 Apr '08

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:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; bh=bmTIgr+EaK8H7vqLqmJtQ9acKRTQQnLI/Dm8IwqM9UI=; b=a3/zLrtVc+1ETGbmRdWXoViv8NIFkww1Hk2pz5nWIEKKzn1++oR6y1q264lh/JIZpa5zPXNHNr6MKclgwa8QHTuWSdsQFWG+7bP9Jqa3GJ0EgXqEC4ml8/qQFGJdHuANNIC8B/6lC2gkF+YBriLao2wC6ORJtD+72IvAvZs+6Nc= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=message-id:date:from:to:subject:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=ojNsfPJu1EnB5C7OUezo96f70Xtp7AjkbeIP1BPhl4yQn/1o7sIbOnvliiGJ6Uk9T3gP5/yUh/oFkhYWP9qrd+s5xeCyG/IzSYIeVpOm5IENSYOc8bVOOpE2TU+mXX66ISGRUTvh7lW1h9ut96+cK7gfBQYorBRv9idXnOYFxf0= I get something like that if I really hammer on my disk drive for several hours. I'll have to try it again to see what the selector is though. Michael Crawford mdcrawford(a)gmail.com http://www.geometricvisions.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

A couple of kext loading/matching issues
by Valentin Slavov 02 Apr '08

02 Apr '08

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:mime-version:content-type:content-transfer-encoding:content-disposition; bh=5MMmwzbQOMVQ6F3Yto6LsiCkey6C1Z40FNE+hNE3mlk=; b=mOaMTDw0Co1oQ1L8BxrezVUdxzqYSqMiPMiPnRSLvqdtpGWhBcUFdKrZzWL9djxoQOEeVnvjWcUZbLNd/Pr71Bg5zpPy2iq/1/mUy/NONU3ok+F8GBXqNc9pJWAuISniR6ysFsyqJrlJpFkbUHbv5VqqCBnb+f9wagDuLBx9KG4= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=message-id:date:from:to:subject:mime-version:content-type:content-transfer-encoding:content-disposition; b=mTGTvZdY3hobZY61Vuo+dJr+AHT2EQYP5P6EWWD7MYulzpDVdIGlLgvTD5UGfx+7oICQnvdFO5RV6I/epEJ7jHOGGuQ/Z3NwGrcCd3UlssZh04jVomy6qmW/ZEokAkjL1/OttIF88IPpW28SMAtGJut4co5fYUfcfUqDWjfPZyc= Hey guys I'm having a hard time figuring out a couple of issues concerning the loading of kexts. 1) I'm developing a driver which operates on LVM/LDM stripes. However, I'm using an internal library which parses and constructs volumes from the metadata on the devices. This implies that I don't want IOFDiskPartitionScheme or IOGUIDPartitionScheme to get in my way. That's why I increased the matching score and my driver is loaded for the ldm/lvm devices instead of the aforementioned two. Now all of this works fine as long as my driver is loaded (that is, I manually kextload'ed it before). I've read that drivers are usually loaded on demand, which if I understand correctly, means that if I put the kext in /System/Library/Extensions and if a nub (or media device in this) is published which my driver matches on it should be loaded automatically. Is this correct? This doesn't seem to be working in my case (IOGUIDPartitionSheme is loaded instead) and my driver can't match the whole media device any more. So I'd like to know how to make my driver load automatically on the ldm/lvm disks instead of the partition scheme drivers. 2) An interesting issue occurs sometimes when I move drivers to the /System/Library/Extensions folders. When a certain kext (it happens with a few third-party kexts) is anywhere but in /S/L/E it gets kextload'ed without problems. However, when I put it there I get "Bus error" on load. Interestingly enough though, it *loads* fine with kextload -t -v. (I've tested this on a macbook pro with a 9.2.0 kernel, however it doesn't work with the new 9.2.2 kernel as well.) 3) Is it possible to load a few drivers for a given nub and if yes how? Any help would be greatly appreciated. Thanks in advance V. _______________________________________________ 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: detecting process termination in kernel?
by Rick Macklem 02 Apr '08

02 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Tue, 1 Apr 2008, Terry Lambert wrote: For the process related stuff (something NFSv4 calls OpenOwners), I also keep track of the process's creation time, so that recycled pids don't cause grief. (I seem to recall getting the process's creation time isn't a kpi call, so it is one of the places my Kext has to "cheat". It's an open source experimental project, so I'm confortable with that.) rick _______________________________________________ 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.… [good stuff snipped] Also, proc_find() can return false positives, so you will not want to poll either; there are a finite number of PIDs available, and eventually they get recycled. If you are trying to do something by PID, then you are probably Good point. It works ok for my case because all I want to do is some garbarge collection cleanup after the process is gone, so false positives simply delay that. I just create a kernel thread that loops around, sleeping for a fraction of a second and then does proc_find(), so the false positive won't occur unless the system is creating lots'a'processes anyhow. Obviously not necessarily the case for other Kexts. This email sent to site_archiver(a)lists.apple.com

1 0

Re: detecting process termination in kernel?
by Andrew Gallatin 02 Apr '08

02 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Terry Lambert writes: > On Mar 30, 2008, at 8:24 PM, Greg wrote: > > Hi, given several process IDs, what are the different ways of > > finding out (being notified, etc), of when a specific PID > > terminates? I know that kqueues can be used in userland (with > > EVFILT PROC), can they be used as well from a KEXT? > > If you are a KEXT, you typically export a device node and have the > process open it. When the process terminates, the resource tracking > close for the device node will be triggered as part of the process > exit, which will notify you that it has gone away. Except, of course, on some versions of MacOSX if you have wired ("prepared") any memory on behalf of the process via a cdevsw ioctl, and it exits with the memory still wired. Then it will be stuck in a zombie state where the cdevsw close is not called, but the process has been removed from the process table, and pfind()/proc_find() can't find it. Drew _______________________________________________ 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: ioctl(_IOR,'t',19,44) messages
by Justin McKillican 02 Apr '08

02 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi, thanks again, -justin Godfrey On 2008-03-31, at 11:32 , Justin McKillican wrote: thanks, -justin Godrey On 2008-03-30, at 22:15 , Justin McKillican wrote: System is: 10.5.2 (x86), Mac Pro, SATA drive on disk0. thanks, -justin _______________________________________________ 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/gvdl%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.… Thanks for getting back to me. I've narrowed it down to a program being run inside Crossover/wine, I guess the confusion is that I've been using same version of Crossover and my application inside Crossover for months and I've only started seeing these messages a week ago. On Apr 2, 2008, at 1:07 AM, Godfrey van der Linden wrote: Somebody is issuing a TIOCGETA against the disk0s2 block storage which needless to say is a no-no. lsof may give you a bit of a hint but it looks to me as if somebody has confused a disk with a tty device, they are not the same thing at all. ttycom.h:#define TIOCGETA _IOR('t', 19, struct termios) /* get termios struct */ I'm not running anything of my own, oddly enough I've noticed this spamming my console (every few seconds) while playing some World of Warcraft. Would you know of a way to track down what's sending those commands? On Mar 31, 2008, at 2:02 PM, Godfrey van der Linden wrote: Some software appears to be sending serial port ioctl's to disk drive 0 and the disk drive is complaining about it. Do you have any of your own software running? I'm getting the following in my console and I'm wondering if someone could help me or point me to the proper list if this isn't the right one: 3/30/08 10:54:55 PM kernel disk0s2: ioctl(_IOR,'t',19,44) is unsupported. 3/30/08 10:54:55 PM kernel disk0s2: ioctl(_IOR,'t',19,44) is unsupported. 3/30/08 10:56:22 PM kernel disk0s2: ioctl(_IOR,'t',19,44) is unsupported. 3/30/08 10:56:22 PM kernel disk0s2: ioctl(_IOR,'t',19,44) is unsupported. I'm trying to determine if this means some sort of hardware failure or driver failure. This email sent to gvdl(a)mac.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: communication from kext to user space application
by Godfrey van der Linden 02 Apr '08

02 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Godfrey On 2008-03-27, at 10:08 , Andrew Gallatin wrote: No. You could create a character device and support read on it, or create an ioctl that copied the data to userspace. Or you could use a sysctl node. Drew _______________________________________________ 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/gvdl%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.… It really depends on what area of the kernel your kext resides in, if it is a standard BSD style kernel module then you are probably publishing through bdevsw/cdevsw. In which case the ioctl route is by far the easiest solution. Implement a simple little ioctl that passes a pointer into your kext, then the kext can copyout the pointer to the client code. There are some details (64/32 bit address spaces, etc) but you can look this up in sample code. If however your code is naturally an IOKit project then the IOKit mechanisms are probably the way to go. As usual on this list there is no right solution it depends on the context of the problem. So i definitely have to use some IOKit stuff even if do not deal with =20= IO at all? I simply need to pass a c-string from my kext to user =20 space. All this stuff seems to be a bit overkill for this task... As you can guess i'm relatively new to this so any useful hint will be =20= This email sent to gvdl(a)mac.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: ioctl(_IOR,'t',19,44) messages
by Godfrey van der Linden 02 Apr '08

02 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Godfrey On 2008-03-31, at 11:32 , Justin McKillican wrote: thanks, -justin Godrey On 2008-03-30, at 22:15 , Justin McKillican wrote: System is: 10.5.2 (x86), Mac Pro, SATA drive on disk0. thanks, -justin _______________________________________________ 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/gvdl%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.… Somebody is issuing a TIOCGETA against the disk0s2 block storage which needless to say is a no-no. lsof may give you a bit of a hint but it looks to me as if somebody has confused a disk with a tty device, they are not the same thing at all. ttycom.h:#define TIOCGETA _IOR('t', 19, struct termios) / * get termios struct */ I'm not running anything of my own, oddly enough I've noticed this spamming my console (every few seconds) while playing some World of Warcraft. Would you know of a way to track down what's sending those commands? On Mar 31, 2008, at 2:02 PM, Godfrey van der Linden wrote: Some software appears to be sending serial port ioctl's to disk drive 0 and the disk drive is complaining about it. Do you have any of your own software running? I'm getting the following in my console and I'm wondering if someone could help me or point me to the proper list if this isn't the right one: 3/30/08 10:54:55 PM kernel disk0s2: ioctl(_IOR,'t',19,44) is unsupported. 3/30/08 10:54:55 PM kernel disk0s2: ioctl(_IOR,'t',19,44) is unsupported. 3/30/08 10:56:22 PM kernel disk0s2: ioctl(_IOR,'t',19,44) is unsupported. 3/30/08 10:56:22 PM kernel disk0s2: ioctl(_IOR,'t',19,44) is unsupported. I'm trying to determine if this means some sort of hardware failure or driver failure. This email sent to gvdl(a)mac.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: detecting process termination in kernel?
by Terry Lambert 02 Apr '08

02 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com -- Terry _______________________________________________ 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 Mar 30, 2008, at 8:24 PM, Greg wrote: Hi, given several process IDs, what are the different ways of finding out (being notified, etc), of when a specific PID terminates? I know that kqueues can be used in userland (with EVFILT PROC), can they be used as well from a KEXT? If you are a KEXT, you typically export a device node and have the process open it. When the process terminates, the resource tracking close for the device node will be triggered as part of the process exit, which will notify you that it has gone away. This is probably not useful for you, since wanting to know about processes other than processes with which you've already established a communications channel generally means you don't really want to solve the problem you are trying to solve by monitoring user space processes. Also, proc_find() can return false positives, so you will not want to poll either; there are a finite number of PIDs available, and eventually they get recycled. If you are trying to do something by PID, then you are probably also not doing what you want to do , since having a PID is not necessarily the same as having a unique identifier for a process. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Which structure does sysctl fill when KERN_FILE parameter is given as input on Mac Os X 10.4.7?
by Terry Lambert 02 Apr '08

02 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Which is why I suggested you popen() lsof instead, and parse its output. -- Terry I cannot see the definition of struct filelist on Mac 10.4.7. Is it one of the private structure that we cannot use ? Hi All, I am currently working on Mac Os X 10.4.7. I m trying to retrieve the list of files opened by a process using sysctl(). The code snippet that I've written is : void* pBuff = NULL ; mib[0] = CTL_KERN; mib[1] = KERN_FILE; size = sizeof(struct filedesc) ; // This is what I tried. This struct is defined in file sys/filedesc.h pBuff = malloc(size) ; while (sysctl((int *)mib, 2, pBuff, &size, NULL, 0) < 0) { if (errno == ENOMEM) { printf("Not sufficient memory allocated\n"); } size *= 2 ; if ((arguments = realloc(pBuff, size)) == NULL) { printf("Realloc error\n"); } printf("Error in getting file struct %s\n and error no is %d\n", strerror(errno),errno) ; } This code compiles and fills the buffer pBuff. Now my question is that the buffer pBuff should to mapped to which structure on Mac 10.4.7?? I have found that it fills struct file or struct xfile on Free BSD. But on Mac Os, I've tried it with struct extern_file and struct filedesc, but both didn't work. Why are the two structs 'file' and 'xfile' not exposed on Mac 10.4.7? If none of the above, then which struct is getting filled by the above sysctl call? I know that there is an alternative way to do the same task, i.e., through lsof. But my project constraints don't let me use that. Also, when I saw the ktrace dump of lsof, I found that it also uses sysctl. Not portably, if you are unwilling to use lsof. We will probably break you every release and potentially every SU. Hope you are OK with that, since you are using undocumented and unsupported SPI by using that sysctl. If you need to know what it packs in the buffer by default on 10.4.7, it's a struct filelist followed by N struct extern_file's. Your best bet when you have questions like these is to look at the xnu sources for the OS version you are interested in, since, as I implied, this can change from version to version of the OS. Unless your constraint is specifically "don't popen lsof", I would recommend using lsof. _______________________________________________ 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.… It's defined in file_internal.h, which is part of the published xnu sources, but which is not published as part of the SDK because we do not support you using sysctl() to do what you are trying to do. On Mar 30, 2008, at 11:00 PM, Manish Chaturvedi wrote: Regards, Manish -----Original Message----- From: darwin-kernel-bounces+manish_chaturvedi=persistent.co.in(a)lists.apple.com [mailto:darwin-kernel-bounces+manish_chaturvedi=persistent.co.in@lists.apple .com] On Behalf Of Terry Lambert Sent: Saturday, March 29, 2008 3:58 AM To: anukriti_shrimal Cc: darwin-kernel(a)lists.apple.com Subject: Re: Which structure does sysctl fill when KERN_FILE parameter is given as input on Mac Os X 10.4.7? On Mar 27, 2008, at 11:38 PM, anukriti_shrimal wrote: Is there some other way to do the same task? -- Terry This email sent to site_archiver(a)lists.apple.com

1 0

Re: Re: Can not copy files >=4GB. Leopard (for NFS)
by Mike Mackovitch 01 Apr '08

01 Apr '08

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com User-agent: Mutt/1.5.17 (2007-11-01) On Tue, Apr 01, 2008 at 11:16:27AM -0700, Doug Rudoff wrote: > > The one NFS server not originally tested was a Mac OS X Server. And > with that, files larger than 4Gb _can_ be copied with the > Finder. Looks like Apple skipped the testing on non-Mac NFS servers. Apple DOES test against non-MacOSX NFS servers. But apparently the copying of files *using the Finder* wasn't as extensively tested. > I compared the NFS pcaps between a 10.4 X Server and a FreeBSD > server. The difference? The X Server's reply to an NFS FSINFO message > sets maxfilesize to UINT64_MAX (0xffffffffffffffff = 16 exabytes), > which is a completely invalid value. > > I modified the nfs server code in the kernel I'm using to use > UINT64_MAX as maxfilesize as well. And now ... the Finder can copy > files more than 4Gb from a 10.5.2 client to my NFS server. > > So, as long as you're willing to modify your kernel, there is a > workaround. The side effect is that the client thinks it can send > files that exceed the NFS server's capabilities. > > This is caused by two bugs on Apple's part: > 1) The X Server NFS server returns an invalid maxfilesize. > 2) 10.5.2's NFS client code will only allow files more than 4Gb to be > transferred if the NFS server's reports its maxfilesize as UINT64_max. Just to be clear... it's NOT the NFS client code that is disallowing the copy. If it was, then copies would not work from the command line either. It's the Finder (and/or the frameworks it uses) that is disallowing the copy because it is checking the following volume capability (taken from getattrlist(2)): VOL_CAP_FMT_2TB_FILESIZE If this bit is set the volume format supports file sizes upto 2TB. This bit does not necessarily mean that the file system does not support file size more than 2TB. This bit does not mean that the currently available space on the volume is 2TB. Introduced with Darwin 8.0 (Mac OS X version 10.4). The NFS client in 10.5 was programmatically setting this based on whether the NFS server reported a maximum file size of 2TB or greater. Makes sense, no? Unfortunately, that capability is not used for what it's documented as being. Instead that capability is being used (by the Finder) as an indication of "large file support" (aka the support of files larger than 4GB). In the end, the fix is in the NFS client code because it needs to be changed to report the value of that capability NOT as it is named/documented, but rather how it is used. > Macko's previous reply indicates that Apple is aware of bug #2 and is > in the process of fixing it for a later release. Oh, and your "bug #1" isn't a bug. The NFSv3 FSINFO.maxfilesize value is an unsigned 64-bit value. UINT64_MAX is a valid unsigned 64-bit value. HTH --macko _______________________________________________ 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