Re: [PATCH] Properly parse uncompressed DNS names
Ping? Am 06.08.2014 um 21:55 schrieb Mathias Hasselmann: Hi, When parsing uncompressed DNS names the end pointer only must be updated when encountering a compressed label, or after parsing all uncompressed labels. Otherwise parse_rr() will interpret the second uncompressed label as resource record header instead of the bits behind the name terminating NUL byte. The attached patch addresses that issue. Thank you, Mathias ___ connman mailing list connman@connman.net https://lists.connman.net/mailman/listinfo/connman -- Join us at Qt Developer Days 2014 - October 6 - 8 at BCC, Berlin Mathias Hasselmann | mathias.hasselm...@kdab.com | Software Engineer KDAB (Deutschland) GmbHCo KG, a KDAB Group company Tel. Germany +49-30-521325470, Sweden (HQ) +46-563-540090 KDAB - Qt Experts - Platform-independent software solutions smime.p7s Description: S/MIME Cryptographic Signature ___ connman mailing list connman@connman.net https://lists.connman.net/mailman/listinfo/connman
Re: [PATCH] Properly parse uncompressed DNS names
On Wed, 2014-08-06 at 21:55 +0200, Mathias Hasselmann wrote: Hi, When parsing uncompressed DNS names the end pointer only must be updated when encountering a compressed label, or after parsing all uncompressed labels. Otherwise parse_rr() will interpret the second uncompressed label as resource record header instead of the bits behind the name terminating NUL byte. The attached patch addresses that issue. Please send patches inline, it's much easier for everybody that way. That said, there is unfortunately something wrong with the patch. With current upstream code I can be pretty confident that 'host www 127.0.0.1' returns the proper name and address of the web server. After testing with your patch it starts to fail with 'host' complaining about malformed packets. I think the patch is doing some use case properly as you would not have sent it otherwise. Can you provide packet traces for loopback and the uplink interface before and after your changes so that we can figure out what kind of packets get delivered properly after the fix? Cheers, Patrik ___ connman mailing list connman@connman.net https://lists.connman.net/mailman/listinfo/connman
Re: [PATCH] Properly parse uncompressed DNS names
Hi Patrik, Sad to hear that the patch doesn't work for you. Have to take another look: Guess while fixing uncompressed labels I've broken the more common compressed variant? Anyway. The case fixed is with a (very primitive) fallback nameserver that answers questions like this: \X53\X0E\X00\X20 \X00\X01\X00\X00\X00\X00\X00\X01 \X07connman\X03net\X00\X00\X01\X00\X01 \X00\X00\X29\X10\X00\X00\X00\X00\X00\X00\X00 or this: \X37\X38\X00\X20 \X00\X01\X00\X00\X00\X00\X00\X01 \X03106\X0211\X03145\X03198\X07in-addr\X04arpa\X00\X00\X0C\X00\X01 \X00\X00\X29\X10\X00\X00\X00\X00\X00\X00\X00 with packets like this: \x53\x0E\x80\xA0 \x00\x01\x00\x01\x00\x00\x00\x00 \x07connman\x03net\x00\x00\x01\x00\x01 \x07connman\x03net\x00\x00\x01\x00\x01 \x00\x00\x07\x08\x00\x04\xC6\x91\x0B\x6A and this: \X37\X38\X80\XA0 \X00\X01\X00\X01\X00\X00\X00\X00 \X03106\X0211\X03145\X03198\X07in-addr\X04arpa\X00\X00\X0C\X00\X01 \X03106\X0211\X03145\X03198\X07in-addr\X04arpa\X00\X00\X0C\X00\X01 \X00\X00\X07\X08\X00\X0D\X07connman\X03net\X00 As you can see the answers' labels are uncompressed instead of just pointing to the questions' labels. Ciao, Mathias Am 12.08.2014 um 12:46 schrieb Patrik Flykt: On Wed, 2014-08-06 at 21:55 +0200, Mathias Hasselmann wrote: Hi, When parsing uncompressed DNS names the end pointer only must be updated when encountering a compressed label, or after parsing all uncompressed labels. Otherwise parse_rr() will interpret the second uncompressed label as resource record header instead of the bits behind the name terminating NUL byte. The attached patch addresses that issue. Please send patches inline, it's much easier for everybody that way. That said, there is unfortunately something wrong with the patch. With current upstream code I can be pretty confident that 'host www 127.0.0.1' returns the proper name and address of the web server. After testing with your patch it starts to fail with 'host' complaining about malformed packets. I think the patch is doing some use case properly as you would not have sent it otherwise. Can you provide packet traces for loopback and the uplink interface before and after your changes so that we can figure out what kind of packets get delivered properly after the fix? Cheers, Patrik ___ connman mailing list connman@connman.net https://lists.connman.net/mailman/listinfo/connman -- Join us at Qt Developer Days 2014 - October 6 - 8 at BCC, Berlin Mathias Hasselmann | mathias.hasselm...@kdab.com | Software Engineer KDAB (Deutschland) GmbHCo KG, a KDAB Group company Tel. Germany +49-30-521325470, Sweden (HQ) +46-563-540090 KDAB - Qt Experts - Platform-independent software solutions smime.p7s Description: S/MIME Cryptographic Signature ___ connman mailing list connman@connman.net https://lists.connman.net/mailman/listinfo/connman
[PATCH] Properly parse uncompressed DNS names
Hi, When parsing uncompressed DNS names the end pointer only must be updated when encountering a compressed label, or after parsing all uncompressed labels. Otherwise parse_rr() will interpret the second uncompressed label as resource record header instead of the bits behind the name terminating NUL byte. The attached patch addresses that issue. Thank you, Mathias -- Join us at Qt Developer Days 2014 - October 6 - 8 at BCC, Berlin Mathias Hasselmann | mathias.hasselm...@kdab.com | Software Engineer KDAB (Deutschland) GmbHCo KG, a KDAB Group company Tel. Germany +49-30-521325470, Sweden (HQ) +46-563-540090 KDAB - Qt Experts - Platform-independent software solutions From ce2bede8c2de445e99262bb66b9effedbf186ced Mon Sep 17 00:00:00 2001 From: Mathias Hasselmann mathias.hasselm...@kdab.com Date: Wed, 6 Aug 2014 21:47:37 +0200 Subject: [PATCH] Properly parse uncompressed DNS names When parsing uncompressed DNS names the end pointer only must be updated when encountering a compressed label, or after parsing all uncompressed labels. Otherwise parse_rr() will interpret the second uncompressed label as resource record header instead of the bits behind the name terminating NUL byte. --- src/dnsproxy.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/dnsproxy.c b/src/dnsproxy.c index 7232b98..2de8bdc 100644 --- a/src/dnsproxy.c +++ b/src/dnsproxy.c @@ -880,14 +880,14 @@ static int get_name(int counter, p += label_len + 1; - if (!*end) -*end = p; - if (p = max) return -ENOBUFS; } } + if (!*end) + *end = p + 1; + return 0; } -- 1.9.1 smime.p7s Description: S/MIME Cryptographic Signature ___ connman mailing list connman@connman.net https://lists.connman.net/mailman/listinfo/connman
