[kvm-unit-tests PATCH 0/2] pci: pci_host_bridge_get_paddr() can not fail

[kvm-unit-tests PATCH 0/2] pci: pci_host_bridge_get_paddr() can not fail

[Alexander Gordeev]

Cc: Thomas Huth <thuth@redhat.com>
Cc: Andrew Jones <drjones@redhat.com>

Alexander Gordeev (2):
  pci: Make all ones invalid translate address
  pci: pci_host_bridge_get_paddr() can not fail

 lib/pci-host-generic.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

-- 
1.8.3.1

[kvm-unit-tests PATCH 1/2] pci: Make all ones invalid translate address

[Alexander Gordeev]

Function pci_host_bridge_get_paddr() returns zero in case
no mapping for a passed PCI address is found. All ones is
a better choice, since zero may be a legitimate physical
address.

Suggested-by: Andrew Jones <drjones@redhat.com>
Cc: Thomas Huth <thuth@redhat.com>
Cc: Andrew Jones <drjones@redhat.com>
Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
---
 lib/pci-host-generic.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/pci-host-generic.c b/lib/pci-host-generic.c
index d57b47ee4c53..7f72d649144e 100644
--- a/lib/pci-host-generic.c
+++ b/lib/pci-host-generic.c
@@ -271,7 +271,7 @@ phys_addr_t pci_host_bridge_get_paddr(u64 pci_addr)
 		as++;
 	}
 
-	return 0;
+	return ~0;
 }
 
 static void __iomem *pci_get_dev_conf(struct pci_host_bridge *host, int devfn)
-- 
1.8.3.1

[kvm-unit-tests PATCH 2/2] pci: pci_host_bridge_get_paddr() can not fail

[Alexander Gordeev]

Function pci_host_bridge_get_paddr() is passed a PCI bus
address to translate into physical one. The PCI address
is always read from a device BAR and is expected to be
valid. Therefore, a failure to find the mapping for the
PCI address indicates a serious implementation problem
elsewhere.

Cc: Thomas Huth <thuth@redhat.com>
Cc: Andrew Jones <drjones@redhat.com>
Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
---
 lib/pci-host-generic.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/lib/pci-host-generic.c b/lib/pci-host-generic.c
index 7f72d649144e..433a0c8e4d15 100644
--- a/lib/pci-host-generic.c
+++ b/lib/pci-host-generic.c
@@ -271,6 +271,7 @@ phys_addr_t pci_host_bridge_get_paddr(u64 pci_addr)
 		as++;
 	}
 
+	assert(0);
 	return ~0;
 }
 
-- 
1.8.3.1

Re: [kvm-unit-tests PATCH 0/2] pci: pci_host_bridge_get_paddr() can not fail

[Andrew Jones]

On Fri, Nov 25, 2016 at 03:56:04PM +0100, Alexander Gordeev wrote:
> Cc: Thomas Huth <thuth@redhat.com>
> Cc: Andrew Jones <drjones@redhat.com>
> 
> Alexander Gordeev (2):
>   pci: Make all ones invalid translate address
>   pci: pci_host_bridge_get_paddr() can not fail
> 
>  lib/pci-host-generic.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> -- 
> 1.8.3.1

I applied the first to arm/next, but not the other.  I think
returning ~0 makes more sense here, and that callers (even
callers that are part of the API) should consider asserting
on receiving it.

As we discussed, low-level API should return whatever it
finds/doesn't find (no asserts). Higher-level API can assert
on invalid inputs and unexpected results from the low-level
API.

Additionally, this series doesn't solve the problem I reported,
which is, if you do

 pci_bar_get_addr(any_device, invalid_bar_num)

you get back the base address of the PIO region. That happens
because pci_bar_get(any_device, invalid_bar_num) returns zero
and then that zero gets translated correctly by pci_translate_addr
to the PIO region's physical base address.

I think we should look into adding asserts to pci_bar_get/set_addr
and the other higher-level API calls instead. We should add

 assert(pci_bar_is_valid(bar_num))

many places, and pci_bar_get_addr should have


 phys_addr = pci_translate_addr(dev, pci_addr)
 assert(phys_addr != ~0)
 
Thanks,
drew