November 2004 - Darwin-kernel

udp_input() and CSUM_PSEUDO_HDR
by Andrew Gallatin 30 Nov '04

30 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi, I have a driver which does TCP and UDP checksum offload. The device tells me the partial (payload) checksum, which I fill into m_pkthdr.csum_data. However, udp_input() ignores partial checksums: /* * Checksum extended UDP header and data. */ if (uh->uh_sum) { if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) { if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR) uh->uh_sum = m->m_pkthdr.csum_data; else goto doudpcksum; uh->uh_sum ^= 0xffff; FreeBSD's udp_input() is mostly identical, except in place of goto doudpcksum; it has the code to integrate the partial checksum: uh->uh_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr, htonl((u_short)len + m->m_pkthdr.csum_data + IPPROTO_UDP)); Is this just a bug, or is it a workaround for a broken device? If the latter, is there any way the workaround could be changed to allow for non-broken drivers to pass partial checksums up the stack? Right now, since I want to handle UDP checksum offload, I have some ugly code in my driver figure out if its a tcp or udp packet, and to call in_pseudo() on it (as above) to get the complete checksum. Its pretty ugly, and adds some cost to the receive path. It would be nice to be able to remove it. Thanks, 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: Licensing Issue
by Kevin Van Vechten 30 Nov '04

30 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi Girish, You may find this article helpful: <http://developer.apple.com/darwin/news/qa20011004.html#License>. - Kevin On Nov 24, 2004, at 8:50 PM, Justin Walker wrote: On Nov 24, 2004, at 20:26, Girish wrote: Dear All, This is from the darwin site (<http://developer.apple.com/darwin>): Regards, 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/site_archiver%40lists.… We are adding the support of SnapShot to HFS+ File system and for achieving this we are modifying the source code of HFS and VFS available with Darwin kernel. We want to know that can we redistribute this as a proprietary solution or is there any Licensing issue involve in this. You (or your lawyers) should read the APSL, which, I think, is pretty clear about the use of software from the Darwin base. Each component available on the darwin site has a license associated with it, and you should be clear about the license associated with each. "Before you start working with open source projects, please read and familiarize yourself with the licensing terms that apply to that project. For access to Apple-developed open source projects, you will need to register as a member of the Apple open source community by accepting the Apple Public Source License and choosing a user name and password. Non-Apple third party open source projects can be accessed without such registration." Check <http://developer.apple.com/darwin/licensing.html>. I think the import of the APSL is that you can't use source covered by APSL in a proprietary way: you have to push your modifications to that source back to Apple and the Darwin community. If you use source from the Darwin site that is covered by other licenses, you have to abide by them. Two ancillary points: - Don't cross-post, please. It is an annoyance for others who may not be subscribed to all the lists, and clutters the mailboxes of those who are. - Don't use HTML when posting to these lists. It may look nice in your mailer, but it likely doesn't in others. -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | If you're not confused, | You're not paying attention *-------------------------------------- *-------------------------------* _______________________________________________ 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/ kevin%40opendarwin.org This email sent to kevin(a)opendarwin.org This email sent to site_archiver(a)lists.apple.com

1 0

Re: Combining mbufs
by Justin Walker 29 Nov '04

29 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Nov 29, 2004, at 13:18, Matt Jaffa wrote: Hi, Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | Some people have a mental | horizon of radius zero, and | call it their point of view. | -- David Hilbert *--------------------------------------*-------------------------------* _______________________________________________ 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.… Is there an easy way to combine mbufs, I have a sosend filter on sockets, and on the same socket I am collecting the mbufs together if they meet a certain condition, and when the last one arrives I would like to put all of these mbufs into one. Why do you want to? Unless the chain gets unreasonably long, the system should be able to handle it with no problems. If you really need to do this, then you need to pay attention to the "shape" of each mbuf (the bits in the flags word) in the chain. Just blindly copying chunks of bits from one place to another (as in your code snippet) won't do what you want. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Combining mbufs
by Andrew Gallatin 29 Nov '04

29 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Matt Jaffa writes: > Hi, > > Is there an easy way to combine mbufs, I have a sosend filter on > sockets, > and on the same socket I am collecting the mbufs together if they meet > a certain condition, > and when the last one arrives I would like to put all of these mbufs > into one. > Depending on exactly what you want, m_dup() may be what you're looking for. See bsd/kern/uipc_mbuf.c 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

Combining mbufs
by Matt Jaffa 29 Nov '04

29 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi, 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.… Is there an easy way to combine mbufs, I have a sosend filter on sockets, and on the same socket I am collecting the mbufs together if they meet a certain condition, and when the last one arrives I would like to put all of these mbufs into one. I can successfully exctract the data out of them, but putting that data all back into 1 mbuf has been giving me loads of problems. I tried to create a new mbuf using MGETHDR, to get the first mbuf as a packet header, then do unsigned char * help = mtod(mbuf,unsigned char*); memcpy(help,ourdatacombined,length); I was looking in Steven's book TCP/IP and darwin sources to get these ideas, but seems to not be working. This email sent to site_archiver(a)lists.apple.com

1 0

Reinject mbuf in Socket NKE
by Matt Jaffa 29 Nov '04

29 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi, 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 been looking over source code and searching the web but don't really come up with anything. I have a socket NKE filter which on some packets returns a EJUSTRETURN so I can send them out later. But for a socket NKE I don't know how to reinject them. I tried sbappend(&so->so_snd, mymbuf); so is the socket from the filtered sosend function. It doesn't seem to be working, I have injected mbufs using the so->so_rcv, and that has worked for putting my own data on the in-buffer. If you go to this webpage: http://developer.apple.com/documentation/Darwin/Conceptual/ NetworkKernelExtensions/about/chapter_1_section_7.html and search the page for reinject, it says that a method for reinjecting is not defined yet, will this ever be defined, or is it defined and I don't know about it. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Panic on Data access error . Why?
by Justin Walker 27 Nov '04

27 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Nov 27, 2004, at 5:05, Parav Pandit wrote: Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | When LuteFisk is outlawed | Only outlaws will have | LuteFisk *--------------------------------------*-------------------------------* _______________________________________________ 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.… Hi, In any IOKit driver, whenever there is any Data Access Error, we get kernel panic. Please let me ask, WHY? Why do we have kernel panic on data access error? In Linux this kind of Segmentation faults at kernel level doesn't panic. Opinions will always vary on this subject, but keep in mind that a segmentation fault means that something is seriously wrong. If the fault occurs in the kernel, it's very difficult to determine, automatically, whether the problem is "containable". The philosophy in Mac OS X is to treat this as a serious problem that should not be happening, and panic as soon as it is detected, to provide an opportunity to determine the source of the problem. This email sent to site_archiver(a)lists.apple.com

1 0

Panic on Data access error . Why?
by Parav Pandit 27 Nov '04

27 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Importance: Normal Hi, In any IOKit driver, whenever there is any Data Access Error, we get kernel panic. Please let me ask, WHY? Why do we have kernel panic on data access error? In Linux this kind of Segmentation faults at kernel level doesn't panic. Is it the PowerPC 970 architecture or is it the kernel? I think we can take care this exception, if we can print the whole back trace on screen. It is true, that such segmentation fault must not be there in kernel code, but there should not be panic. isn't it? Application or driver thread should be terminated on such exceptions. 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.… This email sent to site_archiver(a)lists.apple.com

1 0

IOKit + display models
by Alex Reynolds 26 Nov '04

26 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com User-agent: Internet Messaging Program (IMP) 3.2.2 I'd like to figure out how to get from the IOKit call-equivalent of: [Sond] reynolda$ ioreg -c AppleDisplay | grep DisplayVendorID | | | "DisplayVendorID" = 1552 | | | "DisplayVendorID" = 4268 [Sond] reynolda$ to the corresponding list of the two monitor models connected to my machine, which the Display system preference says are an "Apple Studio Display" and a "Dell 1504FP". The files in /System/Library/Displays/Overrides/DisplayVendorID-610 are not in a consistent format as far as the information they return. I'll bet that isn't where this information is kept anyway. There's an entry in there that matches 1552 to an Apple Studio Display but I can't find anything for non-Apple displays (e.g. matching the 4268 entry to the Dell display). (This table must be *somewhere* if the Display system preference knows it.) Does anyone know how to bend IOKit's ear to return this information? TIA, Alex _______________________________________________ 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: Licensing Issue
by Justin Walker 25 Nov '04

25 Nov '04

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Nov 24, 2004, at 20:26, Girish wrote: Dear All, This is from the darwin site (<http://developer.apple.com/darwin>): Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | If you're not confused, | You're not paying attention *--------------------------------------*-------------------------------* _______________________________________________ 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.… We are adding the support of SnapShot to HFS+ File system and for achieving this we are modifying the source code of HFS and VFS available with Darwin kernel. We want to know that can we redistribute this as a proprietary solution or is there any Licensing issue involve in this. You (or your lawyers) should read the APSL, which, I think, is pretty clear about the use of software from the Darwin base. Each component available on the darwin site has a license associated with it, and you should be clear about the license associated with each. "Before you start working with open source projects, please read and familiarize yourself with the licensing terms that apply to that project. For access to Apple-developed open source projects, you will need to register as a member of the Apple open source community by accepting the Apple Public Source License and choosing a user name and password. Non-Apple third party open source projects can be accessed without such registration." Check <http://developer.apple.com/darwin/licensing.html>. I think the import of the APSL is that you can't use source covered by APSL in a proprietary way: you have to push your modifications to that source back to Apple and the Darwin community. If you use source from the Darwin site that is covered by other licenses, you have to abide by them. Two ancillary points: - Don't cross-post, please. It is an annoyance for others who may not be subscribed to all the lists, and clutters the mailboxes of those who are. - Don't use HTML when posting to these lists. It may look nice in your mailer, but it likely doesn't in others. This email sent to site_archiver(a)lists.apple.com

1 0