Re: [PATCH v4 2/3] lib: logic_pio: Reject accesses to unregistered CPU MMIO regions

From: John Garry
Date: Thu Jun 13 2019 - 11:41:13 EST

On 13/06/2019 04:20, Bjorn Helgaas wrote:
On Tue, Jun 11, 2019 at 10:12:53PM +0800, John Garry wrote:
Currently when accessing logical indirect PIO addresses in
logic_{in, out}{,s}, we first ensure that the region is registered.

Hi Bjorn,

I think logic_pio is specifically concerned with I/O port space, so
it's a little bit unfortunate that we named this "PIO".

PIO is a general term for "Programmed I/O", which just means the CPU
is involved in each transfer, as opposed to DMA. The transfers can be
to either MMIO or I/O port space.

So this ends up being a little confusing because I think you mean
"Port I/O", not "Programmed I/O".

Personally I agree that the naming isn't great. But then Arnd does think that "PIO" is appropriate.

There were many different names along the way to this support merged, and I think that the naming became almost irrelevant in the end.

However, no such check exists for CPU MMIO regions. The CPU MMIO regions
would be registered by the PCI host (when PCI_IOBASE is defined) in

IIUC this "CPU MMIO region" is an MMIO region where a memory load or
store from the CPU is converted by a PCI host bridge into an I/O port
transaction on PCI.


Again IIUC, logic_pio supports two kinds of I/O port space accesses:

1) The simple "bridge converts loads/stores to an MMIO region to PCI
I/O port transactions" kind, and

2) The more complicated "somebody supplies logic_pio_host_ops
functions that do arbitrary magic to generate I/O port
transactions on some bus.


And this patch is making the first kind smarter. Previously it would
perform the memory access whenever "addr < MMIO_UPPER_LIMIT", but
after this patch it will only do it if find_io_range() succeeds.

Right? Sorry for restating what probably should have been obvious to

Yes, right. A logical PIO range is registered for a PCI MMIO region in pci_register_io_range(). As such, if no range is registered, we can assume that no PCI MMIO region has been mapped and discard any attempted access.

I have two observations here:

1) The simple "bridge converts CPU MMIO space to PCI I/O port space"
flavor is essentially identical to what ia64 (and probably other
architectures) does. This should really be combined somehow.

Maybe. For ia64, it seems to have some "platform" versions of IO port accessors, and then also accessors need a fence barrier. I'm not sure how well that would fit with logical PIO. It would need further analysis.

IIRC, PPC did have its own custom version of "indirect IO". So we could look to factor that out.

2) If you made a default set of logic_pio_host_ops that merely did
loads/stores and maybe added a couple fields in the struct
logic_pio_hwaddr, I bet you could unify the two kinds so
logic_inb() would look something like this:

Yeah, I did consider this. We do not provide host operators for PCI MMIO ranges. We could simply provide regular versions of inb et al for this. A small obstacle for this is that we redefine inb et al, so would need "direct" versions also. It would be strange.

So I'm not sure on the value of this.

u8 logic_inb(unsigned long addr)
struct logic_pio_hwaddr *range = find_io_range(addr);

if (!range)
return (u8) ~0;

return (u8) range->ops->in(range->hostdata, addr, sz);

We have seen scenarios when systems which don't have a PCI host, or they
do but the PCI host probe fails, certain drivers attempts to still attempt
to access PCI IO ports; examples are in [1] and [2].

Such is a case on an ARM64 system without a PCI host:

root@(none)$ insmod hwmon/f71805f.ko
Unable to handle kernel paging request at virtual address ffff7dfffee0002e
Mem abort info:
ESR = 0x96000046
Exception class = DABT (current EL), IL = 32 bits
SET = 0, FnV = 0
EA = 0, S1PTW = 0
Data abort info:
ISV = 0, ISS = 0x00000046
CM = 0, WnR = 1
swapper pgtable: 4k pages, 48-bit VAs, pgdp = (____ptrval____)
[ffff7dfffee0002e] pgd=000000000141c003, pud=000000000141d003, pmd=0000000000000000
Internal error: Oops: 96000046 [#1] PREEMPT SMP
Modules linked in: f71805f(+)
CPU: 20 PID: 2736 Comm: insmod Not tainted 5.1.0-rc1-00003-g6f1bfec2a620-dirty #99
Hardware name: Huawei Taishan 2280 /D05, BIOS Hisilicon D05 IT21 Nemo 2.0 RC0 04/18/2018
pstate: 80000005 (Nzcv daif -PAN -UAO)
pc : logic_outb+0x54/0xb8
lr : f71805f_find+0x2c/0x1b8 [f71805f]
sp : ffff000025fbba90
x29: ffff000025fbba90 x28: ffff000008b944d0
x27: ffff000025fbbdf0 x26: 0000000000000100
x25: ffff801f8c270580 x24: ffff000011420000
x23: ffff000025fbbb3e x22: ffff000025fbbb40
x21: ffff000008b991b8 x20: 0000000000000087
x19: 000000000000002e x18: ffffffffffffffff
x17: 0000000000000000 x16: 0000000000000000
x15: ffff00001127d6c8 x14: 0000000000000000
x13: 0000000000000000 x12: 0000000000000000
x11: 0000000000010820 x10: 0000841fdac40000
x9 : 0000000000000001 x8 : 0000000040000000
x7 : 0000000000210d00 x6 : 0000000000000000
x5 : ffff801fb6a46040 x4 : ffff841febeaeda0
x3 : 0000000000ffbffe x2 : ffff000025fbbb40
x1 : ffff7dfffee0002e x0 : ffff7dfffee00000
Process insmod (pid: 2736, stack limit = 0x(____ptrval____))
Call trace:
f71805f_find+0x2c/0x1b8 [f71805f]
f71805f_init+0x38/0xe48 [f71805f]
Code: d2bfdc00 f2cfbfe0 f2ffffe0 8b000021 (39000034)
---[ end trace 10ea80bde051bbfc ]---

Well-behaved drivers call request_{muxed_}region() to grab the IO port
region, but success here still doesn't actually mean that there is some IO
port mapped in this region.

This patch adds a check to ensure that the CPU MMIO region is registered
prior to accessing the PCI IO ports.

Any failed checks silently return.

I *think* what you're doing here is making inb/outb/etc work the same
as on x86, i.e., if no device responds to an inb(), the caller gets
~0, and if no device claims an outb() the data gets dropped.

Correct, but with a caveat: when you say no device responds, this means that - for arm64 case - no PCI MMIO region is mapped.

That should be explicit in the commit log.