On Tue, 23 Aug 2005, Dave North wrote: > Alan Stern wrote: > > > Have you seen this patch? > > http://marc.theaimsgroup.com/?l=linux-usb-devel&m=112439094723976&w=2 > > This started quite a bit of extracurricular activity (I installed the > patch and enumeration started in 2.6.12.3; previously it would _never_ > get that far). This led Alan to ask me to set CONFIG_USB_DEBUG=y, which
Actually I asked you to set CONFIG_USB_STORAGE_DEBUG=y. It's a different, independent setting. > I had unset due to odd behavior when previously trying various kernel > parameters. > At that point I decided to update the kernel again to 2.6.12.5, which > turns out to have caused some confusion. > My first unpatched run with 2.6.12.5 led to two enumeration failures, > so I set about partially applying a newer version of the patch as > requested by Dave Brownell. Enumeration worked erratically with various > values, and I started to get suspicious. So I backed out the patch and > tried 2.6.12.5 all on its lonesome a few more times, and sure enough, > it would enumerate. > More pissed or embarrassed? Hard to say. > Whatever changed between .3 and .5 got it to that state, and adding > lines from the patch made little if any difference. > I should have tested the raw kernel more thoroughly, but as luck would > have it my first tries were complete failures. Now I rarely see > complete failure. > So I have no idea if the patch is useful, at least as regards the > problem I'm seeing. > I did try various other USB2 devices on the port in question, and all > failed, including one another Linux user swears is very reliable. > I do have dmesg output from insertion of the device (16K), an attempt > to mount (16K) and unplugging after failure (another 16K). I also have > the output from /var/log/debug from that session (~540 lines in 62K). > If there is a good reason to post them, I will. If anyone just wants > the files forwarded, please let me know. > My sincere apology for the haphazard testing of 2.6.12.5 giving the > impression the patch (rather than the kernel) was getting through > enumeration. However, the good news is, it _is_ getting through > enumeration most times now. > Still, nothing USB2 mounts on that port unless I use 2.4.x, in which > case anything will. > As always, let me know if I can screw up anything else in a helpful > way... It would be very helpful if you could conduct a series of careful tests of 2.6.12.3 and 2.6.12.5, each both with and without the patch. Which combinations succeed in enumerating and which don't? Try each several times so that individual flukes won't influence the overall results much. It won't hurt if you also include the debugging log from a mount attempt following a successful enumeration, with usb-storage debugging turned on. Alan Stern ------------------------------------------------------- SF.Net email is Sponsored by the Better Software Conference & EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf _______________________________________________ linux-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
