On Mon, Jan 23, 2017 at 02:33:35PM -0600, Bjorn Helgaas wrote: > On Tue, Jan 17, 2017 at 08:56:28AM -0600, Bjorn Helgaas wrote: > > On Tue, Dec 27, 2016 at 05:57:37PM -0600, Bjorn Helgaas wrote: > > > Hi Killian, > > > > > > Thanks for the report (https://bugzilla.kernel.org/show_bug.cgi?id=190861) > > > and all the debugging you've done. Below is a revert of the troublesome > > > commit. Can you test it and verify that it also fixes the problem? > > > > > > I assume Mika is looking at this and will have a better solution soon. > > > But if not, I'll queue this up for v4.10. > > > > Can somebody please summarize the current state of this issue? I > > assume somebody has already posted a better patch that should replace > > this naive revert, but I haven't been following the whole thread. > > This is somewhat frustrating. Is there a better patch than the revert > mentioned below? There was a lot of hullabaloo when I first posted > it, but I haven't seen a good alternative yet. I intended the revert > as a worst-case scenario fix, with the expectation that somebody would > fix the problem or at least avoid it without having to do the revert. > Maybe somebody posted that better fix and I just missed it? I understood that there is a patch here: https://patchwork.freedesktop.org/patch/132478/ that is supposed to fix the issue. I'm waiting Kilian to test it. > >From my perspective (and I have not followed the whole 100 message > thread), the bare bones of the situation are that 006d44e49a25 ("PCI: > Add runtime PM support for PCIe ports") probably reduced power > consumption on some machines. But it also made Kilian's system > unresponsive when locking the screen. > > Given only those assumptions, a revert seems like a reasonable > approach. I understand and agree that we want to save power, but > not at the expense of making systems unresponsive. But even if you revert the runtime PM commit, the same thing happens when the system is suspended. > Maybe 006d44e49a25 actually fixed a functional problem in addition to > saving power? I don't think the changelog mentions anything like > that, but if that's the case, we should certainly consider that. > > We're at -rc5 already, so if we want something other than a revert, > now is the time to propose it. Hmm, runtime PM patches went in for 4.9 IIRC. It is not a regression introduced in v4.10 release cycle so I'm not sure why we are in such hurry here? I understand that the issue should be fixed but not why it should be fixed for v4.10 as it is not a regression introduced by v4.10-rc1. -- 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
