December 2004 - Darwin-kernel

Re: Socket NKE UDP filter
by Matt Jaffa 18 Dec '04

18 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Matt On Dec 17, 2004, at 8:45 PM, John Davidorff Pell wrote: Why aren't you using the firewall? JP On 17 Dec 2004, at 19:36, Matt Jaffa wrote: Hi, (void)thread_funnel_set(network_flock, funnel_state); return 0; } Thanks, Matt This email sent to johnpell(a)mac.com -- - /~\ The ASCII \ / Ribbon Campaign X Help cure HTML Email / \ _______________________________________________ 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.… Because I am writing an application based firewall. With easy to configure rules, and alerts on an outgoing connection you don't have a rule for. I have written a Socket NKE filter that filters UDP traffic, in the send function I want to know what the foreign ip address is and the port number. I have tried a couple of ways, and both don't work: Here is what I have tried: static int tl_send2(struct socket *so, struct sockaddr **nam, struct uio **uio, struct mbuf **m, struct mbuf **control, int *flags, struct kextcb *kp) { boolean_t funnel_state; funnel_state = thread_funnel_set(network_flock, TRUE); printf("UDP: %d\n",current_proc()->p_pid); if(nam && *nam) { struct sockaddr_in * jk = (struct sockaddr_in *)*nam; printf("Here: %s\n", inet_ntoa(jk->sin_addr)); printf("Sorry: %u\n", jk->sin_addr.s_addr); } else { // what to do if it falls in the else clause? printf("Sorry: It is not null jaffa\n"); } I need to know what to do if it falls in the else statement of this if else. _______________________________________________ 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/ johnpell%40mac.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: Socket NKE UDP filter
by John Davidorff Pell 18 Dec '04

18 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Why aren't you using the firewall? JP On 17 Dec 2004, at 19:36, Matt Jaffa wrote: Hi, (void)thread_funnel_set(network_flock, funnel_state); return 0; } Thanks, Matt _______________________________________________ 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/johnpell%40mac.com This email sent to johnpell(a)mac.com -- - /~\ The ASCII \ / Ribbon Campaign X Help cure HTML Email / \ _______________________________________________ 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 I have written a Socket NKE filter that filters UDP traffic, in the send function I want to know what the foreign ip address is and the port number. I have tried a couple of ways, and both don't work: Here is what I have tried: static int tl_send2(struct socket *so, struct sockaddr **nam, struct uio **uio, struct mbuf **m, struct mbuf **control, int *flags, struct kextcb *kp) { boolean_t funnel_state; funnel_state = thread_funnel_set(network_flock, TRUE); printf("UDP: %d\n",current_proc()->p_pid); if(nam && *nam) { struct sockaddr_in * jk = (struct sockaddr_in *)*nam; printf("Here: %s\n", inet_ntoa(jk->sin_addr)); printf("Sorry: %u\n", jk->sin_addr.s_addr); } else { // what to do if it falls in the else clause? printf("Sorry: It is not null jaffa\n"); } I need to know what to do if it falls in the else statement of this if else. smime.p7s

1 0

Socket NKE UDP filter
by Matt Jaffa 18 Dec '04

18 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi, static int tl_send2(struct socket *so, struct sockaddr **nam, struct uio **uio, struct mbuf **m, struct mbuf **control, int *flags, struct kextcb *kp) { boolean_t funnel_state; funnel_state = thread_funnel_set(network_flock, TRUE); printf("UDP: %d\n",current_proc()->p_pid); if(nam && *nam) { struct sockaddr_in * jk = (struct sockaddr_in *)*nam; printf("Here: %s\n", inet_ntoa(jk->sin_addr)); printf("Sorry: %u\n", jk->sin_addr.s_addr); } else { // what to do if it falls in the else clause? printf("Sorry: It is not null jaffa\n"); } (void)thread_funnel_set(network_flock, funnel_state); return 0; } Thanks, Matt _______________________________________________ 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.… I have written a Socket NKE filter that filters UDP traffic, in the send function I want to know what the foreign ip address is and the port number. I have tried a couple of ways, and both don't work: Here is what I have tried: I need to know what to do if it falls in the else statement of this if else. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Common way of communicating with mass storage devices
by Godfrey van der Linden 16 Dec '04

16 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Godfrey On Dec 15, , at 23:30, Yogesh P wrote: Hi, _______________________________________________ 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.… You probably want to go over the the ata-scsi list as they have the domain expertise. However I can only answer your question with a question of my own. The correct answer depends on what you want to do with the API. We already have a common way of talking to all of these devices through the IOMedia/IOBlockStorage APIs. So maybe your work has already been done for you. Finally an observation. If this code is intended for a single application then it is FAR, FAR better to write your 'independant' layer out in user land rather than the kernel, using native APIs. This is a whole lot better to debug and you don't tie up kernel resources unnecessarily. I am working on the code which will provide a common way of communicating with mass storage devices(SCSI, SATA, USB, FireWire etc.). I am trying to implement this module in device service layer, as the document "Writing Drivers for Mass Storage Devices" by Apple specify that the device service layer provides generic block storage driver irrespective of transport layer and storage device underneath. Can any one guide me whether device service layer is the correct place for achieving the common way of communicating with all types of mass storage devices? This email sent to site_archiver(a)lists.apple.com

1 0

Common way of communicating with mass storage devices
by Yogesh P 16 Dec '04

16 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi, I am working on the code which will provide a common way of communicating with mass storage devices(SCSI, SATA, USB, FireWire etc.). I am trying to implement this module in device service layer, as the document "Writing Drivers for Mass Storage Devices" by Apple specify that the device service layer provides generic block storage driver irrespective of transport layer and storage device underneath. Can any one guide me whether device service layer is the correct place for achieving the common way of communicating with all types of mass storage devices? Thanks & Regards, Yogesh Pahilwan _______________________________________________ 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

file last access time late sync
by Alain Bench 15 Dec '04

15 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Mail-followup-to: Alain Bench <albench(a)oreka.com>, Darwin kernel ml <darwin-kernel(a)lists.apple.com> User-agent: Mutt/1.4i-ja.1 Hello, Some people using Darwin 7.6.0 and HFS+ journaled filesystem have a problem where a file's last access time is not updated immediately on a file read. | $ cd ~/Mail | $ ls -ul testbox; cat testbox >/dev/null; date; ls -ul testbox | -rw------- 1 chris chris 943280 10 Dec 02:56 testbox | Fri Dec 10 03:00:47 CET 2004 | -rw------- 1 chris chris 943280 10 Dec 02:56 testbox Atime is updated to the right 03:00 something as about 5 minutes later. Journaling enabled or disabled same problem. Allegedly Darwin 6.x had no problem. Is this late atime a bug? Or is it a setting somewhere? Note: I know nothing about Darwin or MacOS. The problem was found during a report of misbehaviour of the MUA Mutt on comp.mail.mutt, see there full context in thread "N-flag stays put after reading", or <URL:http://www.google.com/groups?threadm=2004-12-01.07-28-53@krille.blacktr…>. Bye! Alain. -- Give your computer's unused idle processor cycles to a scientific goal: The Folding@home project at <URL:http://folding.stanford.edu/>. _______________________________________________ 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: kextload error: "kld(): Undefined symbols: __ZN11IOUSBDevice9metaClassE"
by Dean Reece 14 Dec '04

14 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Cheers, - Dean On Dec 14, 2004, at 2:39 PM, Brian Jorgage wrote: Can anyone shed some light on 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.… Did you change the OSBundleLibraries in you Info.plist to include the USB family? That property is used to determine which kexts your kext must link against. The HelloIOKit example does not need the USB family, so it isn't listed as a dependency. Attempting to kextload a driver when the above error (see subject line) occurs. Driver consists of the basic HelloIOKit example plus first few lines of AppleUSBComposite::start. Specifically this line: "_device=OSDynamicCast(IOUSBDevice, provider)". When this call is absent the driver loads. When present I get the above error. Somehow kextload doesn't recognize the IOUSBDevice symbol used in OSDynamicCast. At least that's what I gather from the kld() error message. I am including the IOUSBDevice.h header in cpp file. Brian Jorgage This email sent to site_archiver(a)lists.apple.com

1 0

kextload error: "kld(): Undefined symbols: __ZN11IOUSBDevice9metaClassE"
by Brian Jorgage 14 Dec '04

14 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Attempting to kextload a driver when the above error (see subject line) occurs. Driver consists of the basic HelloIOKit example plus first few lines of AppleUSBComposite::start. Specifically this line: "_device=OSDynamicCast(IOUSBDevice, provider)". When this call is absent the driver loads. When present I get the above error. Somehow kextload doesn't recognize the IOUSBDevice symbol used in OSDynamicCast. At least that's what I gather from the kld() error message. I am including the IOUSBDevice.h header in cpp file. Can anyone shed some light on this ? Brian Jorgage _______________________________________________ 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: can IOKit driver be paged out?
by Dean Reece 10 Dec '04

10 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Not a silly question - some other OS's allow kernel memory to be paged. Cheers, - Dean Regards, Parav Pandit _______________________________________________ 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 Dec 10, 2004, at 07:28, Parav Pandit wrote: I just want to know, whenever we load IOKit driver using kextload command, does IOKit driver wired down in the memory or can it be paged out whenever there is a lack of memory? and what about the xnu_kernel, I think it can't be paged out otherwise who will do this paging? IOKit drivers, kernel text, and most kernel memory allocations are wired for all versions of OS X. If your kext is really huge, you can break it into pieces that load on demand (have one small kext that monitors the port for an open, and then brings in the "real" driver to handle I/O). If I asked a very silly question, please pardon me for it. If you are interested in tracking your kext's memory usage, here are a few tools you can use: 1) /usr/sbin/kextstat shows you how much memory is wired for text and static data for each loaded kext. 2) /usr/sbin/ioclasscount shows you how many instances of each class you care about are present. 3) /usr/sbin/ioalloccount shows you how much memory is being used by IOKit in total (not counting allocations made via non-IOKit APIs) You can often spot leaks in your code by writing a loop that executes ioclasscount for all classes your driver defines or references, as well as ioalloccount, followed by some major operation for your driver (could be load/unload, could be open/close, etc...). You then watch for any steady ascent in any of the allocation numbers. This email sent to site_archiver(a)lists.apple.com

1 0

Re: can IOKit driver be paged out?
by Garth Cummings 10 Dec '04

10 Dec '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi Parav, On Dec 10, 2004, at 7:28 AM, Parav Pandit wrote: Hi, Regards, Parav Pandit All kernel memory is wired by default and isn't paged. This is a good reason to use kernel memory sparingly. --gc __________________________________________________________________ Garth Cummings Apple Developer Technical Support gcummings(a)apple.com http://developer.apple.com/technicalsupport _______________________________________________ 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 I just want to know, whenever we load IOKit driver using kextload command, does IOKit driver wired down in the memory or can it be paged out whenever there is a lack of memory? and what about the xnu_kernel, I think it can't be paged out otherwise who will do this paging? If I asked a very silly question, please pardon me for it. smime.p7s

1 0