Re: [PATCH v41 11/24] x86/sgx: Add SGX misc driver interface

[Dave Hansen]

On 11/14/20 8:01 PM, Hillf Danton wrote:
> On Fri, 13 Nov 2020 00:01:22 +0200 Jarkko Sakkinen wrote:
> +
> +static unsigned long sgx_get_unmapped_area(struct file *file,
> +					   unsigned long addr,
> +					   unsigned long len,
> +					   unsigned long pgoff,
> +					   unsigned long flags)
> +{
> +	if ((flags & MAP_TYPE) == MAP_PRIVATE)
> +		return -EINVAL;
> +
> +	if (flags & MAP_FIXED)
> +		return addr;
> +
> MAP_FIXED is checked in arch_get_unmapped_area() along with others,
> so better to check it always there.

In the whole tangle of get_unmapped_area() functions, it wasn't patently
obvious to me that MAP_FIXED functionality is guaranteed to be present
in the higher-level functions.

There are also a fair number of these fops->get_unmapped_area()
instances that have the same MAP_FIXED line.  Are in need of fixups too?
 Or are you suggesting that since SGX is x86-only, it can depend on the
x86 arch_get_unmapped_area() code to continue to have this MAP_FIXED
functionality?