May 2005 - Darwin-dev - lists.apple.com

Re: bind() not working for outbound socket connections...
by Philip George 22 May '05

22 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On May 22, 2005, at 11:40 AM, Vincent Lubet wrote: [1] if i add a route, how do i make sure that the new route is chosen and nothing else? gotcha. makes perfect sense. [2] after using it, does the route need to be removed explicitly, or is closing the socket i used thru it enough? that's what i thought. just needed to hear it in plain english. [3] does adding a new route or deleting a route only apply to the context of my app or does it affect the entire system? that's what i thought. just needed to hear it in plain english. [4] i have already played with this a little, but the examples i'm going off of are a little over-the-top. if you know of a simple example of how this is done in C, please post a link. in any case, you're post was very helpful. thanks for answering all of my questions. - philip _______________________________________________ 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… The routing engine picks the route that is the more specific for the destination address. A host route is the more specific kind of route, so simply add a host route with the IP address of the dead router. You need to remove the route yourself, there is no connection between the socket layer and the routing table. It's a system wide setting (that's why it's a privileged operation) I suggest to have a look at the source code of the "route" command in Darwin or FreeBSD repository. that's actually one of the sources i was referring to above. but, since my last post, i've cleared out most of the stuff i'm not interested in and I can now see the some semblance of simplicity in it. and by the way, Giuliano, thanks for actually helping out instead of talking down to me and then giving me a reading assignment. i've been plagued with unhelpful responses on this issue on three lists... mostly from the same guy chasing me from list to list. Because you have out of the ordinary requirements you really need to understand what you are doing especially when changing system wide settings. understood. but, it's presumptuous to assume that i don't already get that. it's not helpful to avoid answering questions and instead repeatedly blast back negativity and reading assignments at someone. the posts i'm referring to were not done in a spirit of helpfulness. they were done in a spirit of know-it-all-ism. not cool. This email sent to site_archiver(a)lists.apple.com

1 0

Re: bind() not working for outbound socket connections...
by Vincent Lubet 22 May '05

22 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On May 22, 2005, at 3:56 AM, Philip George wrote: On May 22, 2005, at 3:43 AM, Giuliano Gavazzi wrote: now THIS is the kind of thinking it's gonna take to get this working! great suggestion, and creative. i have a couple of questions about this suggestion: [1] if i add a route, how do i make sure that the new route is chosen and nothing else? [2] after using it, does the route need to be removed explicitly, or is closing the socket i used thru it enough? [3] does adding a new route or deleting a route only apply to the context of my app or does it affect the entire system? It's a system wide setting (that's why it's a privileged operation) [4] i have already played with this a little, but the examples i'm going off of are a little over-the-top. if you know of a simple example of how this is done in C, please post a link. Vincent Giuliano, you rock. - philip _______________________________________________ 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/vlubet%40apple.com This email sent to vlubet(a)apple.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… well, would it not be enough to add manually a route for the particular remote ip/net going through the dead router (and thus the interface he wants)? it requires that my app will have to be set to run as root unfortunately, but at this point i'm desperate. plus, i've buttoned up all possible places where buffers are flooded with user input. everything's bounds-checked. The routing engine picks the route that is the more specific for the destination address. A host route is the more specific kind of route, so simply add a host route with the IP address of the dead router. You need to remove the route yourself, there is no connection between the socket layer and the routing table. I suggest to have a look at the source code of the "route" command in Darwin or FreeBSD repository. and by the way, Giuliano, thanks for actually helping out instead of talking down to me and then giving me a reading assignment. i've been plagued with unhelpful responses on this issue on three lists... mostly from the same guy chasing me from list to list. Because you have out of the ordinary requirements you really need to understand what you are doing especially when changing system wide settings. This email sent to site_archiver(a)lists.apple.com

1 0

Re: bind() not working for outbound socket connections...
by Philip George 22 May '05

22 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On May 22, 2005, at 3:43 AM, Giuliano Gavazzi wrote: now THIS is the kind of thinking it's gonna take to get this working! great suggestion, and creative. i have a couple of questions about this suggestion: [1] if i add a route, how do i make sure that the new route is chosen and nothing else? [2] after using it, does the route need to be removed explicitly, or is closing the socket i used thru it enough? [3] does adding a new route or deleting a route only apply to the context of my app or does it affect the entire system? [4] i have already played with this a little, but the examples i'm going off of are a little over-the-top. if you know of a simple example of how this is done in C, please post a link. Giuliano, you rock. - philip _______________________________________________ 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… well, would it not be enough to add manually a route for the particular remote ip/net going through the dead router (and thus the interface he wants)? it requires that my app will have to be set to run as root unfortunately, but at this point i'm desperate. plus, i've buttoned up all possible places where buffers are flooded with user input. everything's bounds-checked. and by the way, Giuliano, thanks for actually helping out instead of talking down to me and then giving me a reading assignment. i've been plagued with unhelpful responses on this issue on three lists... mostly from the same guy chasing me from list to list. This email sent to site_archiver(a)lists.apple.com

1 0

Re: bind() not working for outbound socket connections...
by Giuliano Gavazzi 22 May '05

22 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Giuliano _______________________________________________ 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 6:54 pm -0700 2005/05/21, Justin Walker wrote: On May 21, 2005, at 18:28, Philip George wrote: i have a system with two ports, one is connected to a router that CAN see the internet and and one is connected to a router that CANNOT see the internet. if i try to bind() a socket to the port that connects to the "dead" router, the system still autoroutes it through the good interface to get the message to the web server out on the internet. for most practical purposes, that feature rocks, but for my particular needs, i have to be sure that the messages are being routed out through a specific interface. if there is no good route to the destination address thru the chosen interface, then the subsequent connect() should fail (as opposed to looking for another route). i **don't** want these particular messages autorouted thru any random interface. You really do need to read enough to understand IP routing. You can complain all you want, but the kernel routing engine is still going to pick an outbound interface that takes your packets in the direction of the destination address. The interfaces aren't chosen randomly; they are chosen in a deterministic way by the routing engine. well, would it not be enough to add manually a route for the particular remote ip/net going through the dead router (and thus the interface he wants)? This email sent to site_archiver(a)lists.apple.com

1 0

passwordserver_sasl-59.2
by Dale Walsh 22 May '05

22 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Mta-interface: amavisd-new-2.3.1-rc3-E (2005-05-06) at daleenterprise.com Does anyone know anything about this? -- Dale _______________________________________________ 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… A question was posed to me regarding the fix for libsql.so and the poser wanted to know if it was going to be included in the next passwordserver_sasl update/release and I didn't know what to say. This email sent to site_archiver(a)lists.apple.com

1 0

Re: all HFS+ CLIs
by Martin Costabel 22 May '05

22 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Mozilla Thunderbird 1.0 (Macintosh/20041206) Peter Seebach wrote: [] /System/Library/CoreServices/FixupResourceForks dir -- Martin _______________________________________________ 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 am also dreading what happens when a user makes a "safe" tarball on a Tiger machine and opens it on a Panther machine... You will get the files split in AppleDouble format, that is if the file "file" had a resource fork, it will be represented in the tarball as 2 files "file" (without resource fork) and "._file". Unpacking on Panther or any other non-Tiger system will give you these 2 files. You can then run on any directory "dir" produced by the unpacking. This will give you more or less the same result as unpacking on Tiger. In both cases, the file creation dates on the files with resource forks will be wrong: they will correspond to the unpacking/Fixup times, not to the creation times of the original. This email sent to site_archiver(a)lists.apple.com

1 0

Re: all HFS+ CLIs
by seebs＠plethora.net 22 May '05

22 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <Pine.BSO.4.62.0505222333070.29110(a)invert.10k.org>, jake writes: >The webpage says that >cp, mv, tar, rsync are now HFS/resource forks aware. >is that a complete list? I seem to recall hearing that scp has an option for it. I am not sure whether or not there's a complete list. I am also dreading what happens when a user makes a "safe" tarball on a Tiger machine and opens it on a Panther machine... I do not feel entirely confident that the boundary conditions will be handled gracefully. -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: accessing prom
by Dmitry Markman 22 May '05

22 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Laurence Harris from Carbon list made some comments { Str255 computerSerialNumber; CFTypeRef cfComputerSerialNumber = NULL; io_object_t ioEntry; io_iterator_t ioIterator; kern_return_t kernResult; mach_port_t masterPort; bzero(computerSerialNumber,sizeof(computerSerialNumber)); kernResult = IOMasterPort(MACH_PORT_NULL, &masterPort); On May 21, 2005, at 11:21 PM, Justin Walker wrote: On May 21, 2005, at 18:12, Dmitry Markman wrote: thank you very much Justin the following code does the job De nada! Could be; elegance is in the eye of the coder, of course... Cheers, Justin -- Justin C. Walker, Curmudgeon-At-Large Institute for General Semantics -------- "Weaseling out of things is what separates us from the animals. Well, except the weasel." - Homer J Simpson -------- _______________________________________________ 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/dmarkman%40mac.com This email sent to dmarkman(a)mac.com Dmitry Markman _______________________________________________ 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… so here is the modified code: (specifically I replaced CFStringGetCStringPtr to CFStringGetCString also I replaced kCFStringEncodingMacRoman by CFStringGetSystemEncoding ()) if(kernResult == KERN_SUCCESS) { kernResult = IORegistryCreateIterator (masterPort,kIOServicePlane,0,&ioIterator); } if(kernResult == KERN_SUCCESS) { ioEntry = IOIteratorNext(ioIterator); if(ioEntry != 0) { cfComputerSerialNumber = IORegistryEntrySearchCFProperty(ioEntry,kIOServicePlane,CFSTR (kIOPlatformSerialNumberKey),kCFAllocatorDefault,0); IOObjectRelease(ioEntry); } IOObjectRelease(ioIterator); } if(cfComputerSerialNumber != NULL) { if(CFStringGetTypeID() == CFGetTypeID (cfComputerSerialNumber)) { Boolean err = CFStringGetCString((CFStringRef) cfComputerSerialNumber,(char *)computerSerialNumber,sizeof (computerSerialNumber),CFStringGetSystemEncoding()); if(!err) bzero(computerSerialNumber,sizeof (computerSerialNumber)); } CFRelease(cfComputerSerialNumber); } printf("S/N: %s\n",(char *)computerSerialNumber); } however I suppose that there is a more elegant way by using some matching facility This email sent to site_archiver(a)lists.apple.com

1 0

Re: accessing prom
by Justin Walker 22 May '05

22 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On May 21, 2005, at 18:12, Dmitry Markman wrote: thank you very much Justin De nada! Could be; elegance is in the eye of the coder, of course... Cheers, Justin -- Justin C. Walker, Curmudgeon-At-Large Institute for General Semantics -------- "Weaseling out of things is what separates us from the animals. Well, except the weasel." - Homer J Simpson -------- _______________________________________________ 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… the following code does the job however I suppose that there is a more elegant way by using some matching facility This email sent to site_archiver(a)lists.apple.com

1 0

Re: bind() not working for outbound socket connections...
by Justin Walker 22 May '05

22 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On May 21, 2005, at 18:28, Philip George wrote: Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large Institute for General Semantics -------- "Weaseling out of things is what separates us from the animals. Well, except the weasel." - Homer J Simpson -------- _______________________________________________ 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… why is bind() not letting me specify a particular interface (on a multihomed system) to send socket messages thru? Because that is not bind()'s job. bind() does one thing: it tells the kernel to assign a network address, which you provide, to a local endpoint, specified by the value that you get from a socket() call. The network address consists, for PF_INET, an IPv4 address and a port; you can specify either, both, or neither, in your call. If you specify neither, your endpoint will be assigned a port number by the kernel. If you do not specify an address, it will be effectively set to INADDR_ANY (i.e., "*"). If you specify an address, that address will be used for outbound packets; if you do not, the kernel will pick one based on the destination address. If you chose an address that does not "go towards the Internet", and you don't have routing enabled, then your packet won't go anywhere. This is as close as you can get to binding to an interface: you can choose an address that is associated to the interface you are interested in. i have a system with two ports, one is connected to a router that CAN see the internet and and one is connected to a router that CANNOT see the internet. if i try to bind() a socket to the port that connects to the "dead" router, the system still autoroutes it through the good interface to get the message to the web server out on the internet. for most practical purposes, that feature rocks, but for my particular needs, i have to be sure that the messages are being routed out through a specific interface. if there is no good route to the destination address thru the chosen interface, then the subsequent connect() should fail (as opposed to looking for another route). i **don't** want these particular messages autorouted thru any random interface. You really do need to read enough to understand IP routing. You can complain all you want, but the kernel routing engine is still going to pick an outbound interface that takes your packets in the direction of the destination address. The interfaces aren't chosen randomly; they are chosen in a deterministic way by the routing engine. This email sent to site_archiver(a)lists.apple.com

1 0