Hi! On Saturday 14 September 2002 01:34, Jeff Garzik wrote: > Gernot Hillier wrote: > > You can find all details and the patch in the archives (search for > > "epic100.c fiber support, improved") or on the net on > > > > http://www.hillier.de/linux/epic100-fiber.php3. > > Your patch doesn't simply add fiber support, it also sprinkles in a > couple udelay(200)'s and changes the media init code for non-fiber > cards. I want to study it in-depth and test it locally to assure myself > it doesn't break anything. The udelay(200)'s are unfortunately necessary for some cards to initialize the MII successful. We had many "MII not ready" errors spit out by the orginal driver which were fixed with these udelay's for the SMC9432FTX. They were BTW taken out from the linux driver SMC provides for its cards (http://www.smc-europe.com/english/support/driver_manual/adapter/download/9432/9432linux24.zip) But don't waste your time with these - they are not really nice to read. I also have a report from a tester that my patch fixes some problems with a non fiber card which didn't work before. As it's not 100% clear and I cannot reproduce this here I don't claim these udelay's fixes a real bug, though. Two of these three udelay's are also absolutely necessary for the fiber interface to initialize reliably here. > A major change to the media init code should be carefully studied in > 2.4.x series, and is not something to just throw at Marcelo once it just > works for you... If there's anything I can do to help you with this, please let me know. We tried to be as careful as possible to not break anything with the old media init code and to keep our changes as minimal as possible. The last version of our patch does test for presence of LXT970 transceiver and does enable the new code only if one was found. -- Ciao, Gernot - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
