[OpenAFS-devel] klog with openafs-snap-2004-08-16 in SuSE-9.1
(linux 2.6.5-7.104)
Derek Atkins
warlord@MIT.EDU
Sat, 21 Aug 2004 15:11:52 -0400
This first patch to sock_create() implies the configure test is
wrong when trying to determine whether sock_create has a fifth
argument or not. Try fixing the configure test, because I know
the gatekeepers wont accept this patch as-is.
-derek
Sven Menke <Menke@MPPMU.MPG.DE> writes:
> Hi,
>
> I've managed to get the client and klog up and running under the stock
> SuSE 9.1 kernel with the following 2 patches to the August-16
> snapshot:
>
> --- src/rx/LINUX/rx_knet-original.c 2004-07-29 04:46:48.000000000 +0200
> +++ src/rx/LINUX/rx_knet.c 2004-08-21 09:17:22.000000000 +0200
> @@ -38,7 +38,7 @@
>
>
> #ifdef LINUX_KERNEL_IS_SELINUX
> - code = sock_create(AF_INET, SOCK_DGRAM, IPPROTO_UDP, &sockp, 0);
> + code = sock_create(AF_INET, SOCK_DGRAM, IPPROTO_UDP, &sockp);
> #else
> code = sock_create(AF_INET, SOCK_DGRAM, IPPROTO_UDP, &sockp);
> #endif
> --- src/kauth/authclient-orig.c 2004-08-11 17:45:38.000000000 +0200
> +++ src/kauth/authclient.c 2004-08-21 12:14:07.885214536 +0200
> @@ -527,13 +527,13 @@
> oanswer.SeqLen = 0;
> oanswer.SeqBody = (char *)&answer;
>
> - version = 2;
> - code =
> - kawrap_ubik_Call(KAA_AuthenticateV2, conn, 0, name, instance, start,
> - end, &arequest, &oanswer);
> - if (code == RXGEN_OPCODE) {
> - oanswer.MaxSeqLen = sizeof(answer);
> - oanswer.SeqBody = (char *)&answer;
> +// version = 2;
> +// code =
> +// kawrap_ubik_Call(KAA_AuthenticateV2, conn, 0, name, instance, start,
> +// end, &arequest, &oanswer);
> +// if (code == RXGEN_OPCODE) {
> +// oanswer.MaxSeqLen = sizeof(answer);
> +// oanswer.SeqBody = (char *)&answer;
> version = 1;
> code =
> ubik_Call(KAA_Authenticate, conn, 0, name, instance, start, end,
> @@ -550,7 +550,7 @@
> if (code == RXGEN_OPCODE) {
> code = KAOLDINTERFACE;
> }
> - }
> +// }
> if (code) {
> UNLOCK_GLOBAL_MUTEX if ((code >= KAMINERROR) && (code <= KAMAXERROR))
> return code;
>
> I don't know why kawrap_ubik_Call fails for me, I just observed that
> without the patch it overwrites some memory (e.g. ans_label) and klog
> falls back to version 0 authentication and eventually returns:
>
> Unable to authenticate to AFS because AuthServer returned incorrect response
>
> With the patches above klog works.
>
> Cheers,
> --Sven
>
--
Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
Member, MIT Student Information Processing Board (SIPB)
URL: http://web.mit.edu/warlord/ PP-ASEL-IA N1NWH
warlord@MIT.EDU PGP key available