linux-pci.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* IORESOURCE_PCI_FIXES & __pci_bus_assign_resources()
@ 2019-06-17  8:35 Benjamin Herrenschmidt
  0 siblings, 0 replies; only message in thread
From: Benjamin Herrenschmidt @ 2019-06-17  8:35 UTC (permalink / raw)
  To: linux-pci; +Cc: Bjorn Helgaas, Richard Henderson, Matt Turner

Hi !

While working on consolidating resource assignment, I stumbled upon
something alpha does (and maybe others I haven't spotted yet):

On most platforms (not all), it uses the usual pair:

	pci_bus_size_bridges(bus);
	pci_bus_assign_resources(bus);

to reassign everything.

However, before doing so, it first calls pcibios_claim_one_bus() which,
on those platforms (ie those who don't set PCI_PROBE_ONLY), will
effectively claim only resources that have IORESOURCE_PCI_FIXED.

Now, let's leave alone for now the fact that this is will really only
work if those resources aren't behind a bridge as, to the best of my
undertanding of the code at this point, we aren't going to take them
into account when sizing & locating bridges. But that's not my point
right now :-)

From what I can tell, these days, pci_bus_assign_resources() will
already claim those fixed resources via pdev_assign_fixed_resources().

However, it does so *after* it has assigned and claimed resoures for
all the sibling devices on that bus.

That looks wrong to me. Shouldn't we claim the fixed resources first ?
IE something like:

--- a/drivers/pci/setup-bus.c
+++ b/drivers/pci/setup-bus.c
@@ -1345,11 +1345,12 @@ void __pci_bus_assign_resources(const struct pci_bus *bus,
 	struct pci_bus *b;
 	struct pci_dev *dev;
 
+	list_for_each_entry(dev, &bus->devices, bus_list)
+		pdev_assign_fixed_resources(dev);
+
 	pbus_assign_resources_sorted(bus, realloc_head, fail_head);
 
 	list_for_each_entry(dev, &bus->devices, bus_list) {
-		pdev_assign_fixed_resources(dev);
-
 		b = dev->subordinate;
 		if (!b)
 			continue;
?

Now, I suspect most of the time it happens to work due to the fact that
the fixed resources are generally IO resources in the legacy low ranges
( < 0x1000) and PCIBIOS_MIN_IO is *generally* set to 0x1000 but it still
sounds fishy to me.

I don't think I have HW at hand with that type of fixed stuff to test
with at the moment, so this is very much academic right now but I worry
that when I convert archs such as alpha who does that claiming before
the rest of the assignment, switching doing things the other way around
will break.

Any thoughts ?

Cheers,
Ben.
 


^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2019-06-17  8:35 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-06-17  8:35 IORESOURCE_PCI_FIXES & __pci_bus_assign_resources() Benjamin Herrenschmidt

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).