On Wed, Apr 25, 2018 at 08:56:02AM -0700, Omar Sandoval wrote:
> On Tue, Apr 24, 2018 at 03:41:46PM -0600, Keith Busch wrote:
> > +_test_hotplug_slot() {
>
> I'd call this _test_dev_in_hotplug_slot().
Sounds good.
> > + parent="$(_get_pci_parent_from_blkdev)"
>
> I haven't been consistent about asking people to do this, but could you
> make these variables local? I.e.,
>
> local parent
> parent="$(_get_pci_parent_from_blkdev)"
> local slt_cap
> slt_cap=...
No problem.
> > + setpci -s ${parent} CAP_EXP+10.w=10:10
> > + sleep 10
> > + setpci -s ${parent} CAP_EXP+10.w=00:10
>
> For the sake of people of me who don't speak PCI, what do each of these
> commands do? :)
:) Will add to the change log.
This command finds the PCI Express Capability register of the slot the
device is in, then at offset 0x10 (the Link Control Register) writes a
1 to bit 4 (Link Disable). This is happening unbeknownst to any of the
drivers, just like a surprise removal. If this is a capable slot, the
drivers will find out about this through the pcie hotplug handler.
> Should we make the fio job --time_based instead of using
> --size so that we're sure it runs long enough for the sleep?
Right, that makes sense.
Will resend with the fix-ups, but probably not until tomorrow.
