Andy Shevchenko <[email protected]> writes:

> On Tue, May 14, 2019 at 12:23 PM Andy Shevchenko
> <[email protected]> wrote:
>> On Tue, May 14, 2019 at 10:24 AM Esben Haabendal <[email protected]> wrote:
>
>> > Please take a look at https://lkml.org/lkml/2019/4/9/576
>> > ("[PATCH v2 2/4] mfd: ioc3: Add driver for SGI IOC3 chip")
>>
>> Thank you for this link.
>> Now, look at this comment:
>>
>> + /*
>> + * Map all IOC3 registers.  These are shared between subdevices
>> + * so the main IOC3 module manages them.
>> + */
>>
>> Is it your case? Can we see the code?
>
> They do not request resources by the way.

Actually, that looks like a bug in ioc3.c driver.

It is using mfd_add_devices() with a mem_base that has not been properly
requested, and the platform_get_resource() calls made by child drivers
does not guarantee exclusive access to the memory resources, as they are
not inserted in the root memory resource tree.

> You may do the same, I told you this several times.

In drivers/mfd/ioc3.c:

First, the uart resources are defined.  The register memory resource is
defined relative to the mfd driver memory resource.

+static struct resource ioc3_uarta_resources[] = {
+       DEFINE_RES_MEM(offsetof(struct ioc3, sregs.uarta),
+                      sizeof_field(struct ioc3, sregs.uarta)),
+       DEFINE_RES_IRQ(6)
+};

This is then used when creating the uart cell.

+               cell->name = "ioc3-serial8250";
+               cell->id = ioc3_serial_id++;
+               cell->resources = ioc3_uarta_resources;
+               cell->num_resources = ARRAY_SIZE(ioc3_uarta_resources);

Finally, the mfd_add_devices() call is made, giving the resource for the
BAR0 region (&ipd->pdev->resource[0]) as mem_base argument:

+       mfd_add_devices(&ipd->pdev->dev, -1, ioc3_mfd_cells,
+                       cell - ioc3_mfd_cells, &ipd->pdev->resource[0],
+                       0, ipd->domain);

This is just what I want to do.

But in order to guarantee exclusive access to the memory resource, I
need to have it requested.

/Esben

Reply via email to