Date: Wed, 19 Jan 2005 14:01:11 +0100 (CET) From: Barry Bouwsma <freebsd-misuser@remove-NOSPAM-to-reply.NOSPAM.dyndns.dk> To: freebsd-usb@freebsd.org Subject: usb stack merges from wherever... Message-ID: <200501191301.j0JD1A10000721@Mail.NOSPAM.DynDNS.dK>
next in thread | raw e-mail | index | archive | help
Hi Julian or whoever; Is there work-in-progress to merge USB goodies into 4.x, and/or from NetBSD into -current? Since I'm running RELENG_4, I'd be happy to submit patches needed to merge -current usb source into 4.x that I use. Especially if 4.x is lacking. In particular, I'm using current-ified versions of umass.c, and now also ums.c, both of which appear to offer benefits over the 4.x sources. The ums.c code would be easier to plug in as-is. There needs to be one change made to a different file, and then my hacked ums.c should work in both -current and 4.x. I've built new virginal core usb.ko as a module, to make sure I don't have any hidden hacks that make MFC'ed code work -- apart from usbdevs additions that I'm not trying to keep track of -- and I appear to see a slight improvement with -current ums.c over that in 4.x -- the latter would often result in random mouse movements on my screen, which I'm not noticing with -current source. Nothing is obviously broken for me. The files can be found at https://NOSPAM.dyndns.dk/hacks/usb/ums/ (if you can't reach it when you read this, try over the weekend) You'll need to apply the patch for usbhid.h . If there haven't been any changes to -current's ums.c over the past week or two (my CVS repository isn't up-to-date, and it's always good to verify), then the ums.c file at the above URL can be diff'ed against both -current and 4.x to see what I've done. I don't know what would be a candidate for -current for diff reduction, and what would be only applicable for 4.x, so I'll leave that to someone with experience in MFCing or MT4ing. As usual, once my hacks have been reviewed and cleaned up (everything I've done should be found by searching `XXX' and/or `HACK'), you should test and make sure I haven't overlooked anything (like usbdevs), before committing to RELENG_4. My machine is too slow for me to do this in a timely manner. I also would like to submit my MFC/MT4 for umass.c, but this is a bit more intrusive, and also probably requires a few commits to HEAD for new functionality first. It fixes significantly more issues for me. Another thing I've just noticed by using a virginal base usb.ko module instead of what I had been using, is that now I'm having uaudio isoc TD alloc problems again, that had disappeared with my hacked code. I don't know how much of my hacked code is taken from Ian Dowse's megapatches, or if my Works-For-Me hacks are candidates for 4.x if they aren't applicable to -current. I'll see what I can do, if there's interest to fix problems in 4.x. thanks barry bouwsma
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200501191301.j0JD1A10000721>