Hi Bjorn, This is Wei and I work from Facebook Storage Hardware group. The series of changes made by Keith are to support the PCIe hot plug testing on our Lightning platform (Thanks Keith!). Lightning is a “Just of Bunch of Flash”, or JBOF, storage platform that has a PCIe switch connecting to 15 NVMe SSDs. One of our goals is to be able to surprise hot remove/plug each SSDs in the chassis. These changes are to fix issues found during surprise hot plug testing. Specially, they reduced PCIe registers access during a surprised hot remove event. We have incorporated Keith’s patches and tested on our Lightning platform. They worked well. For a detail description of the Lightning platform, please refer to https://code.facebook.com/posts/989638804458007/introducing-lightning-a-flexible-nvme-jbof/ Hope this gives you some background info on these changes. Appreciate your reviews as well! Thanks! -Wei On 11/23/16, 8:09 AM, "Bjorn Helgaas" <helgaas@xxxxxxxxxx> wrote: On Fri, Nov 18, 2016 at 06:25:44PM -0500, Keith Busch wrote: > On Fri, Oct 28, 2016 at 06:58:14PM -0400, Keith Busch wrote: > > Here's version 3. Two of the patches from v2 have already been applied > > upstream, so this is down to three patches now. > > > > The main difference with v2 is setting subordinates to removed when > > detected (patch 1/3). > > > > Patch 2/3 short-cuts pci_device_is_present if is_removed is set. > > > > Otherwise, the same as before. Just to restate the justification, this > > significantly speeds up many device removals without relying on hardware. > > > > Keith Busch (3): > > pci: Add is_removed state > > pci: No config access for removed devices > > pci/msix: Skip disabling removed devices > > > > drivers/pci/hotplug/pciehp_pci.c | 5 +++++ > > drivers/pci/msi.c | 7 ++++++- > > drivers/pci/pci.c | 2 ++ > > drivers/pci/pcie/pcie-dpc.c | 4 ++++ > > 5 files changed, 42 insertions(+), 1 deletion(-) > > Hi Bjorn, > > Just wanted to check in with you on this series. Any thoughts on > this? I've been pinged from several hardware vendors on this since they > say it improves system stability. I keep telling them to bug you about > it, but I guess they think I'm less intimidating. :) I'd think they'd want somebody *more* intimidating to encourage progress :) Sorry I haven't had a chance to look at these yet. I want to think about them a little more because it seems like these should be optimizations, not really fixes. If they improve stability by fixing Linux issues, details of those issues would help. But maybe the improvement is from avoiding things the hardware doesn't handle quite correctly. And I'm always hesitant about things like flags that say "this device is being removed but hasn't been removed yet" because the remove process is already complicated and we've tripped over races and ordering issues in the past. That's not to say this is a bad approach; it's just an excuse for why I need some time to work through these patches. Bjorn ��.n��������+%������w��{.n�����{���"�)�� jg��� �����ݢj����G���� ���j:+v���w�m������w�������h�����٥
