info-vax@ucbvax.UUCP (02/19/86)
In Berkeley's 4.2.? implementation if TCP/IP, there seems to be a hack whereby if the high order bit of the host number in a class B or class A address is set, then such address will be interpreted as a class C or B address, respectively. In other words, if you have a class B address you can have it interpreted as 32766 hosts on one network and 126 subnetworks of 126 hosts apiece (all 0's and all 1's omitted). It looks to me like the code that does this needs to be enabled by haveing the IFF_LOCAL (0x100) flag set in the interface. Now, we don't have the sources for Wollengong's TCP/IP, but we do have the include files. They seem to be the same as the ones for whatever version of Berkeley's Unix that allows this; the definitions are there. This leads me to suspect that if we can get the IFF_LOCAL flag set in interfaces, then we will be able to have subnets with our class B address. Note: this only works for classes A and B, not class C. Does anyone know about this? Is there a way to turn this flag on? If so, would it work? Has anybody actually done it? Would Wollengong talk to us if we did it? (Not as if they seem to be that willing to talk to us anyway, needle, needle). We're not sure we would do it even if we could, but we would like to know if it's possible. -- Eric Norman UUCP: ...{allegra,ihnp4,seismo}!uwvax!uwmacc!ejnorman Pony Express: 1210 West Dayton Street, Madison, WI 53706 Life: Detroit!Alexandria!Omaha!Indianapolis!Madison!Hyde "Tis far easier for a peacock to show his true colors than it is for a lion to swallow his pride." -- Arte Johnson --
info-vax@ucbvax.UUCP (02/20/86)
This is a question about VMS and Wollengong's TCP-IP, not Unix. In Berkeley's 4.2.? implementation if TCP/IP, there seems to be a hack whereby if the high order bit of the host number in a class B or class A address is set, then such address will be interpreted as a class C or B address, respectively. In other words, if you have a class B address you can have it interpreted as 32766 hosts on one network and 126 subnetworks of 126 hosts apiece (all 0's and all 1's omitted). It looks to me like the code that does this needs to be enabled by haveing the IFF_LOCAL (0x100) flag set in the interface. Now, we don't have the sources for Wollengong's TCP/IP, but we do have the include files. They seem to be the same as the ones for whatever version of Berkeley's Unix that allows this; the definitions are there. This leads me to suspect that if we can get the IFF_LOCAL flag set in interfaces, then we will be able to have subnets with our class B address. Note: this only works for classes A and B, not class C. Does anyone know about this. Is there a way to turn this flag on? If so, would it work? Has anybody actually done it? Would Wollengong talk to us if we did it? (Not as if they seem to be that willing to talk to us anyway, needle, needle). We're not sure we would do it even if we could, but we would like to know if it's possible. -- Eric Norman UUCP: ...{allegra,ihnp4,seismo}!uwvax!uwmacc!ejnorman Pony Express: 1210 West Dayton Street, Madison, WI 53706 Life: Detroit!Alexandria!Omaha!Indianapolis!Madison!Hyde "Liberte, Egalite, Homologie." -- Michael Henle --