April 2006 - Darwin-kernel - lists.apple.com

Re: PE_get_hotkey
by Terry Lambert 07 Apr '06

07 Apr '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Apr 6, 2006, at 2:19 PM, Eric Long wrote: There's PE_get_hotkey... As long as you link your KEXTs vs. the correct symbol set: System6.0.exports:_PE_get_hotkey. I asked because this page says kexts should not call pexpert APIs: <http://developer.apple.com/documentation/Darwin/Conceptual/KEXTConcept/KEXT ConceptDependencies/kext_dependencies.html> Your dependency would be on the System6.0 exports, *NOT* pexpert proper. -- 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.… Table 6 Header directories and corresponding kernel subcomponents It's probably more or less correct tha you shouldn't use the APIs, but you were asking how to do something that requires using them - and that's the way you do it. This email sent to site_archiver(a)lists.apple.com

1 0

Re: PE_get_hotkey
by Eric Long 06 Apr '06

06 Apr '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thread-index: AcZZv9L8EWCTqsWzEdq8fAARJH2wOA== Thread-topic: PE_get_hotkey User-agent: Microsoft-Entourage/11.2.1.051004 > There's PE_get_hotkey... > > > As long as you link your KEXTs vs. the correct symbol set: > System6.0.exports:_PE_get_hotkey. I asked because this page says kexts should not call pexpert APIs: <http://developer.apple.com/documentation/Darwin/Conceptual/KEXTConcept/KEXT ConceptDependencies/kext_dependencies.html> > Table 6 Header directories and corresponding kernel subcomponents Eric _______________________________________________ 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: PE_get_hotkey
by Terry Lambert 06 Apr '06

06 Apr '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Apr 5, 2006, at 1:16 PM, Eric Long wrote: Hi, Is there an equivalent to PE_get_hotkey() (in pexpert.h in the kernel framework) that is allowed to be called by kexts? There's PE_get_hotkey... -- 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.… As long as you link your KEXTs vs. the correct symbol set: System6.0.exports:_PE_get_hotkey. This email sent to site_archiver(a)lists.apple.com

1 0

Re: AirPort: out of low-memory clusters
by Ken Mankoff 06 Apr '06

06 Apr '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Apr 6 01:06:39 gouda kernel[0]: AirPort: out of low-memory clusters Apr 6 01:06:39 gouda kernel[0]: AirPort: out of low-memory clusters Apr 6 01:06:39 gouda kernel[0]: AirPort: out of low-memory clusters -k. _______________________________________________ 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.… Sorry to start a new thread on the same topic, I cannot access the computer with my previous emails right now to reply-to. http://lists.apple.com/archives/Darwin-kernel/2006/Mar/msg00091.html Just writing to state that this problem has re-appeared (bug #4498651): I'm leaving the ethernet plug in for now rather than trying to fix the airport just in case someone wants to suggest I look at something specific while the computer is in this error state. This email sent to site_archiver(a)lists.apple.com

1 0

PE_get_hotkey
by Eric Long 05 Apr '06

05 Apr '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thread-index: AcZY7dPEEmGaHMThEdq7ygARJH2wOA== Thread-topic: PE_get_hotkey User-agent: Microsoft-Entourage/11.2.1.051004 Hi, Is there an equivalent to PE_get_hotkey() (in pexpert.h in the kernel framework) that is allowed to be called by kexts? Thanks in advance, Eric _______________________________________________ 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: Modify IP address / Port with a interface filter extension
by Stephane Sudre 05 Apr '06

05 Apr '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On 4 avr. 06, at 21:39, devmaillists wrote: Hello dear mailing list, I like to modify a Ethernet packet with a interface filter extension. The subject of change is the ip destination and port number. For the IP checksum, if you need it: u_int16_t ComputeIPChecksum(u_int16_t * buffer,u_int16_t inLength) { u_int32_t tChecksum=0; if ((inLength & 1) == 0) inLength = inLength >> 1; else inLength = (inLength >> 1) + 1; while (inLength > 0) { tChecksum += ntohs(*(buffer)); buffer += 1; inLength--; } tChecksum = (tChecksum >> 16) + (tChecksum & 0xffff); tChecksum += (tChecksum >> 16); return(~tChecksum); } [...] struct ip * tIPHeader; [...] _______________________________________________ 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 am not very familiar with the mbuf structures but I am currently learning. So here come my questions. Do I have to extract the data of an mbuf int a separate buffer to work on? Or is looping through the mbuf until I find the right position the best way. If I modifiy a byte in the IP data, the CS of the IP Packet and of the Ethernet packet do have to be recalculated or only the IP CS? Is there a call doing the recalculation or is it best to recalc it by my own? Does anybody have s short sample? tIPHeader->ip_sum=0; tIPHeader->ip_sum=htons(ComputeIPChecksum((u_int16_t *) tIPHeader,tIPHeader->ip_hl*4)); This email sent to site_archiver(a)lists.apple.com

1 0

Re: Modify IP address / Port with a interface filter extension
by Josh Graessley 04 Apr '06

04 Apr '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Apr 4, 2006, at 12:39 PM, devmaillists wrote: Hello dear mailing list, I like to modify a Ethernet packet with a interface filter extension. The subject of change is the ip destination and port number. Afraid not. -josh _______________________________________________ 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 am not very familiar with the mbuf structures but I am currently learning. So here come my questions. Do I have to extract the data of an mbuf int a separate buffer to work on? Or is looping through the mbuf until I find the right position the best way. You can walk through the mbuf chain. You might run in to trouble if the data you are interested in is spread across two mbufs. There is an mbuf_pullup function to make the data contiguous, but it mail fail under certain conditions and cause the mbuf chain to be released. If I modifiy a byte in the IP data, the CS of the IP Packet and of the Ethernet packet do have to be recalculated or only the IP CS? This is tricky because the checksum is sometimes computed in hardware and sometimes it is done in software. For the hardware, the stack calculates a small portion of the checksum based on the source and destination ip address and protocol I think. My memory of these bits is vague, so some of the details might be off. Anyhow, this pre- computed value is stuff in to a part of the mbuf for hardware that requires it. To avoid all the possible different problems you could run in to, you should call mbuf_outbound_finalize first. This will force the checksum to be calculated. Then modify the header as you see fit and update the checksums as appropriate. Is there a call doing the recalculation or is it best to recalc it by my own? You will have to calculate the new checksum on your own. There is no KPI for doing this. There is a bug requesting a KPI to force the calculation of the checksum. Does anybody have s short sample? smime.p7s

1 0

Modify IP address / Port with a interface filter extension
by devmaillists 04 Apr '06

04 Apr '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hello dear mailing list, I like to modify a Ethernet packet with a interface filter extension. The subject of change is the ip destination and port number. Does anybody have s short sample? Any tips welcome. Thank you very much in advance. Jürgen kodierwerk(a)mac.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.… I am not very familiar with the mbuf structures but I am currently learning. So here come my questions. Do I have to extract the data of an mbuf int a separate buffer to work on? Or is looping through the mbuf until I find the right position the best way. If I modifiy a byte in the IP data, the CS of the IP Packet and of the Ethernet packet do have to be recalculated or only the IP CS? Is there a call doing the recalculation or is it best to recalc it by my own? Szenario: several Macs in a Network. One Mac shares its network to the outside with the other Macs trough Internetsharing. iMac1 types www.something.com in its safari. It will see a page located at: www.auth.com and has to type a password. Next time the Gateway Mac knows the MACADDr of iMac1 and allows the connection to the outside. This email sent to site_archiver(a)lists.apple.com

1 0

Re: software firewall wishlist
by Terry Lambert 04 Apr '06

04 Apr '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi, I wish pf or ipfilter be ported to darwin so we will have choices for better firewalls (ipfw, pf, ipf) You might want to try "man ipfw"; you are running Tiger, right? 8-). -- 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 Apr 1, 2006, at 12:52 AM, Jett Tayer wrote: OT: ipfilter now runs on linux For pf and ipfilter, either one could be ported as an NKE (Network Kernel Extension) pretty trivially, and Apple provides NKE sample code already. This email sent to site_archiver(a)lists.apple.com

1 0

Re: spin locks and SMP xnu?
by Michael Smith 03 Apr '06

03 Apr '06

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Apr 2, 2006, at 11:47 PM, Parav Pandit wrote: Help me to understand the xnu SMP kernel. Effectively, yes. Effectively, yes. = 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.… 1. In SMP xnu environement on Power MAC G5 system, both the processors share the same kernel address space, because we have only one copy of kernel symbol table. 2. Does it mean that both the processors will have the same page table or TLB entries for kernel space code (kernel, drivers, filesystem)? 3. Each processor will run its own schedular but both of them will share the same task queue to pick up the task to execute. Is that correct? 4. We compile linux kernel for uniprocessor or SMP enviornment. This uni or SMP decide the code for the spin locks. Either disable/ enable interrupt for uniprocessor or Do the spinning and locking operation in addition to disabling inetrrupts. We don't do this compliation stuff for the xnu kernel; how xnu kernel takes care of this code optimization and functionality ? Unlike Linux, the xnu kernel is designed to support a single loadable module binary regardless of the number of CPUs. All of these details are taken care of for you - simply use the most appropriate synchronisation primitive, and assume that it will be optimised accordingly. This email sent to site_archiver(a)lists.apple.com

1 0