--- On Sun, 1/9/11, Martin Fuzzey <mfuzzey@xxxxxxxxx> wrote: usb@xxxxxxxxxxxxxxx > Date: Sunday, January 9, 2011, 4:37 AM > David Brownell wrote: > It'd help shake out bugs > > appearing in some newer HCDs. > > > > > No it's not currently in Greg's queue. > > There was some discussion between you, Alan and myself > about whether it is better to add extra tests, I can live with extra numbered tests in the kernel test driver. add > extra checks > to the existing tests or add parameters to the existing tests. but would rather have no more parameters (they'd cause needless confusion). > > I think Greg is waiting for a consensus on this before > taking > the patch. > > > > Similar comment for peripheral drivers, with gadget > zero... > > > > p.s. ISTR you also had some thoughts about getting the > test script > > into the kernel tree too. Ideally we'd have not > just a bash script > > but also the operator instructions ... but for now, > I'd be glad just > > to see drivers that demand word alignment start to > fail in lab tests > > rather than in the field... > > > > > Yes my opinion is that we should: > 1) Add extra explicit tests for alignment problems (what > this patch does) OK, another numbered test ... > > I think this is superior to merging with the existing tests > because it > will give a clear indication of where the problem lies (one > test will > pass and another will fail if HCDs require alignment) > > 2) Put the test script in the kernel tree (to make it more > likely that the > new tests are called) > > 3) Agreed that some documentation / instructions would be > good. > However I'm not sure what this should really contain and > the > relation to the materiel available at http://www.linux-usb.org/usbtest/ It'll suffice, IMO, to reference the tests in the kernel tree. After all, most Linux testing is done using tools and instructions external to the kernel tree. Offsiting like linux-usb.org site is the norm > > If we agree about this I'll submit patches for 1 + 2. I > think 3 needs > more discussion. The please sumit (a) a patch adding a testcase for at least host-side unaligned-transfers. And if you like, a usbtest driver calling that and the current testcases. For #3, I suspect the best solution requires the Linux Foundation to put test plans in place that don't ignore drivers the way most have so far done. -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
