Description of problem: Incoming packets are malformed using an ADMtek AN983B onboard NIC with the (auto-configured) tulip network driver. Version-Release number of selected component (if applicable): Kernel 2.4.22-1.2188.nptl x86_64 Linux Tulip driver version 0.9.15-pre12 (Aug 9, 2002) How reproducible: Network based applications are not working. Steps to Reproduce: 1. Configure device for DHCP, start network. or 2. Ping some host with -s > 30. or Try to use your favorite network application. Actual results: 1. DHCP fails with error "dhclient: 5 bad udp checksums in 5 packets". and 2. Ping fails with error "wrong data byte #30". (it's always #30) and Other network appliations simply dont work. Expected results: No errors. Additional info: Quick work-around: Use the old_tulip driver instead. Files: 1. dhcp.txt: DHCP client error messages from /var/log/messages. 2. ping.txt Output of Ping. 3. capture.eth Capture of DHCP requests and other network traffic. 4. lspci.txt PCI information about the NIC.
Created attachment 99929 [details] DHCP client error messages from /var/log/messages.
Created attachment 99930 [details] Output of Ping.
Created attachment 99931 [details] Capture of DHCP requests and other network traffic.
Created attachment 99932 [details] PCI information about the NIC.
This issue applies to Fedora Core 2 Test 3, Kernel 2.6.5-1.327, Tulip driver version 1.1.13 (May 11, 2002), too. Unfortunately there is no tulip_old available to quickly get around the problem. While comparing the tulip sources with tulip_old ones and playing around with some of the changes, I found out that the following change seem to work for me: # diff -u tulip.original/tulip_core.c tulip/tulip_core.c --- tulip.original/tulip_core.c 2004-04-18 10:51:59.000000000 +0200 +++ tulip/tulip_core.c 2004-05-05 22:25:59.125854056 +0200 @@ -88,7 +88,7 @@ ToDo: Non-Intel setting could be better. */ -#if defined(__alpha__) || defined(__ia64__) || defined(__x86_64__) +#if defined(__alpha__) || defined(__ia64__) static int csr0 = 0x01A00000 | 0xE000; #elif defined(__i386__) || defined(__powerpc__) static int csr0 = 0x01A00000 | 0x8000; Yet I didn't do much testing, but at first sight it looks good. At least I can browse the Web and write this comment through FC2test3. Jan
Thanks for the bug report. However, Red Hat no longer maintains this version of the product. Please upgrade to the latest version and open a new bug if the problem persists. The Fedora Legacy project (http://fedoralegacy.org/) maintains some older releases, and if you believe this bug is interesting to them, please report the problem in the bug tracker at: http://bugzilla.fedora.us/