Hi All, Thanks for your help, Duncan and Giles.
Thanks mainly to Giles' comment, I tracked this down to a USB controller fault which is documented here: http://sourceforge.net/mailarchive/message.php?msg_id=17450397 This patch is for the older kernel 2.6.9, however, on examining the changelog I realised that a similar fix has been incorporated into the brand new 2.6.18: --- Author: Alan Stern <[EMAIL PROTECTED]> Date: Fri May 12 11:41:59 2006 -0400 [PATCH] UHCI: Work around old Intel bug Some old Intel UHCI controllers have a bug that has shown up in a few systems (the PIIX3 "Neptune" chip set). Until now there has not been any simple way to work around the bug, but the lastest changes in uhci-hcd have made it easy. This patch (as684) adds the work-around. --- So I build a 2.6.18 kernel. Apart from a few minor problems caused by me failing to configure the kernel correctly and my second monitor dying (which was VERY annoying at this particular time), it now works. As suggested in the analysis cited above, this fault means that downloading a file over HTTP consisting of all ones (i.e. 0xff * lots) reproduces the fault reliably. I've tested this on the newer kernel and it now downloads correctly. Mark Liste de diffusion modem ALCATEL SpeedTouch USB Pour se désinscrire : mailto:[EMAIL PROTECTED]
