On Sat, Aug 15, 2015 at 3:33 PM, Jay <MyMailClone@xxxxxxxxxxx> wrote: > Am Samstag, 15. August 2015, 09:39:24 schrieb Bjorn Helgaas: >> Hi Jay, >> >> On Sat, Aug 15, 2015 at 7:49 AM, Jay <MyMailClone@xxxxxxxxxxx> wrote: >> > Am Freitag, 14. August 2015, 15:17:52 schrieb Bjorn Helgaas: >> > ... >> > >> >> Since you didn't say anything like that, I assume the patch in comment >> >> #72 of bug #81551 (https://bugzilla.kernel.org/attachment.cgi?id=156301) >> >> would work as well. That has the advantage that it wouldn't penalize >> >> non-storage controllers. >> > >> > The problem was introduced with 3.15 and the kernel is almost at 4.2 now. >> > >> > A general solution for the JMicron-AHCI/PATA-controllers is still missing >> > although available since late September 2014. It was tested by Barto. >> > >> > Seems what I wrote in comment #76 wasn't completely wrong: >> > https://bugzilla.kernel.org/show_bug.cgi?id=81551#c76 >> > >> > So this may be an opportunity to reconsider the approach to solving >> > things like this. >> >> I don't understand your point, except to acknowledge that we are >> imperfect, have limited resources, and make many mistakes in >> diagnosing problems and communicating solutions. Do you have a >> proposal for a better approach? >> >> The patch that Barto tested in late September 2014 >> (https://bugzilla.kernel.org/show_bug.cgi?id=81551#c66) is exactly the >> patch from comment #72 that I mentioned as possibly being a better >> solution. >> >> That patch wouldn't involve me at all, since it doesn't touch PCI. >> Zhang is proposing a PCI change, so I'm asking for a clear changelog. >> >> A changelog is a "write-once, read-many" situation. It's very >> important that it be concise and clear, and it's worth having the >> expert spend extra time writing the log to make it easier for the many >> novices that will read it in the future. > > please don't feel offended. Nobody should. No need to. And I like what you > wrote, it's definitely constructive. > > My point is that instead of looking for a "perfect" solution, a more pragmatic > approach may be more effective in solving things like this. > > And this was exactly what I suggested in comment #76: "Let's be pragmatic, > take this patch and solve the problem now. And then you (developers) may take > your time to look for a better or even the "perfect" solution." > > That way the case would have been closed a year ago. The only problem with that approach is that the owner of the issue now considers the issue closed, so the better solution never happens. Even if it *does* happen, we now have a fix for a fix, which makes it that much more difficult to follow the history. This is just a generic issue with the way Linux works: maintainers have zero leverage after accepting a patch, so anything they really care about has to happen before that. Bjorn -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
