Hello!
I'm writing a device driver for a subdevice of a PCIe multifunctional
device. It's a platform device that has its resources described in the
top-level driver for the multifuctional device.
I followed the common pattern when a memory resource is obtained by
calling platform_get_resource() and then reserved and remapped by
calling devm_ioremap_resource().
res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
base = devm_ioremap_resource(&pdev->dev, res);
The problem with that approach is that the resource appears twice in
/proc/iomem:
fdc60000-fdc6000f : My Dev
fdc60000-fdc6000f : My Dev
I see that devm_ioremap_resource() takes the start address and the
size of the existing resource and reserves the region at the same
address.
Some drivers (e.g. drivers/mfd/db8500-prcmu.c) use devm_ioremap()
instead. But my concern is that the resource is not marked as busy in
that case.
Which approach is preferred? If there a good way to mark an existing
resource as busy (short of direct flag manipulation)? Would it make
sense to have an API to reserve an existing resource?
--
Regards,
Pavel Roskin
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
