On Thu, Oct 20, 2011 at 8:41 PM, simo <idra@xxxxxxxxx> wrote: > On Thu, 2011-10-20 at 16:01 -0500, Shirish Pargaonkar wrote: >> On Thu, Oct 20, 2011 at 3:55 PM, Oskar Liljeblad <oskar@xxxxxxxxxxx> wrote: >> > On Thursday, October 20, 2011 at 22:13, Oskar Liljeblad wrote: >> >> I like the idea of opaque passwords as well, but the problem is that the >> >> password needs to be in UCS-2 for smb. The password is passed through >> >> _my_mbstowcs which converts the 8-bit password to 16-bit UCS-2 simply >> >> by setting high byte to zero. So that makes opaque 8-bit passwords >> >> impossible. One alternative would be to assume passwords in UTF-8 >> >> regardless of iocharset. >> > >> > On another note, it wouldn't surprise me if Microsoft assumes UTF-16 rather >> > than UCS-2 in their CIFS implementation today... >> > >> > Oskar >> > >> > >> >> codepage to unicode conversion does not come into picture during >> authentication. codepage to unicode conversion comes into picture >> once you are sending ntlm response blob over. >> I do not know if ntlm response on machines with different charsets >> can have that charset specific characters e.g. whether ntlm >> response blob on a machine with German character set can >> have an umlaut as part of ntlm response. > > Passwords are UTF-16, I do not think it makes any sense in 2011 to > support anything else but UTF-8 to be honest. > If you are using codepages on the filesystem for some legacy reason that > should still not change the fact the kernel should get a password in > UTF-8. If your charset does not allow you to type the correct charcter > for your password you are screwed anyways it doesn't matter if before or > after the conversion to URF-16 > > Simo. Shirish's subsequent explanation made sense. Also Pavel took a look at this too. Merged. -- Thanks, Steve -- To unsubscribe from this list: send the line "unsubscribe linux-cifs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
