Description of problem: Please consider upgrading current e1000 v2.1.13-k1 to the latest from Intel. /* Change Log * * 5.2.20 9/30/03 * o Bug fix: SERDES devices might be connected to a back-plane * switch that doesn't support auto-neg, so add the capability * to force 1000/Full. * o Bug fix: Flow control settings for hi/lo watermark didn't * consider changes in the Rx FIFO size, which could occur with * Jumbo Frames or with the reduced FIFO in 82547. * o Better propagation of error codes. [Janice Girouard * (janiceg.com)]. * o Bug fix: hang under heavy Tx stress when running out of Tx * descriptors; wasn't clearing context descriptor when backing * out of send because of no-resource condition. * * 5.2.16 8/8/03 * o Added support for new controllers: 82545GM, 82546GB, 82541/7_B1 * o Bug fix: reset h/w before first EEPROM read because we don't know * who may have been messing with the device before we got there. * [Dave Johnson (ddj -a-t- cascv.brown.edu)] * o Bug fix: read the correct work from EEPROM to detect programmed * WoL settings. * o Bug fix: TSO would hang if space left in FIFO was being miscalculated * when mss dropped without a correspoding drop in the DMA buffer size. * o ASF for Fiber nics isn't supported. * o Bug fix: Workaround added for potential hang with 82544 running in * PCI-X if send buffer terminates within an evenly-aligned dword. * o Feature: Add support for ethtool flow control setting. * o Feature: Add support for ethtool TSO setting. * o Feature: Increase default Tx Descriptor count to 1024 for >= 82544. * * 5.1.13 5/28/03 */ Thanks :)
Created attachment 96017 [details] e1000 5.2.20 inserted into the kernel tree Diff from 2.4.22-1.2115.nptl to upgrade e1000 to 5.2.20
Created attachment 96018 [details] diff with spec file changes Diff from 2.4.22-1.2115.nptl kernel-2.4.spec
Guess this one is unnecessary as 2.6 comes with 5.2.20-k1 already. Resolving as RAWHIDE.