Re: [PATCH] mm: Create the new vm_fault_t type

[Souptick Joarder]

On Mon, Nov 5, 2018 at 2:43 PM Mike Rapoport <rppt@xxxxxxxxxxxxx> wrote:
>
> On Mon, Nov 05, 2018 at 11:14:17AM +0530, Souptick Joarder wrote:
> > Hi Matthew,
> >
> > On Sun, Nov 4, 2018 at 2:06 PM Mike Rapoport <rppt@xxxxxxxxxxxxx> wrote:
> > >
> > > On Sat, Nov 03, 2018 at 05:02:36AM -0700, Matthew Wilcox wrote:
> > > > On Sat, Nov 03, 2018 at 10:35:04AM +0530, Souptick Joarder wrote:
> > > > > Page fault handlers are supposed to return VM_FAULT codes,
> > > > > but some drivers/file systems mistakenly return error
> > > > > numbers. Now that all drivers/file systems have been converted
> > > > > to use the vm_fault_t return type, change the type definition
> > > > > to no longer be compatible with 'int'. By making it an unsigned
> > > > > int, the function prototype becomes incompatible with a function
> > > > > which returns int. Sparse will detect any attempts to return a
> > > > > value which is not a VM_FAULT code.
> > > >
> > > >
> > > > > -/* Encode hstate index for a hwpoisoned large page */
> > > > > -#define VM_FAULT_SET_HINDEX(x) ((x) << 12)
> > > > > -#define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf)
> > > > ...
> > > > > +/* Encode hstate index for a hwpoisoned large page */
> > > > > +#define VM_FAULT_SET_HINDEX(x) ((__force vm_fault_t)((x) << 16))
> > > > > +#define VM_FAULT_GET_HINDEX(x) (((x) >> 16) & 0xf)
> > > >
> > > > I think it's important to mention in the changelog that these values
> > > > have been changed to avoid conflicts with other VM_FAULT codes.
> > > >
> > > > > +/**
> > > > > + * typedef vm_fault_t -  __bitwise unsigned int
> > > > > + *
> > > > > + * vm_fault_t is the new unsigned int type to return VM_FAULT
> > > > > + * code by page fault handlers of drivers/file systems. Now if
> > > > > + * any page fault handlers returns non VM_FAULT code instead
> > > > > + * of VM_FAULT code, it will be a mismatch with function
> > > > > + * prototype and sparse will detect it.
> > > > > + */
> > > >
> > > > The first line should be what the typedef *means*, not repeat the
> > > > compiler's definition.  The rest of the description should be information
> > > > for someone coming to the type for the first time; what you've written
> > > > here is changelog material.
> > > >
> > > > /**
> > > >  * typedef vm_fault_t - Return type for page fault handlers.
> > > >  *
> > > >  * Page fault handlers return a bitmask of %VM_FAULT values.
> > > >  */
> > > >
> > > > > +typedef __bitwise unsigned int vm_fault_t;
> > > > > +
> > > > > +/**
> > > > > + * enum - VM_FAULT code
> > > >
> > > > Can you document an anonymous enum?  I've never tried.  Did you run this
> > > > through 'make htmldocs'?
> > >
> > > You cannot document an anonymous enum.
> >
> >
> > I assume, you are pointing to Document folder and I don't know if this
> > enum need to be documented or not.
>
> The enum should be documented, even if it's documentation is (yet) not
> linked anywhere in the Documentation/
>
> > I didn't run 'make htmldocs' as there is no document related changes.
>
> You can verify that kernel-doc can parse your documentation by running
>
> scripts/kernel-doc -none -v <filename>

I run "scripts/kernel-doc -none -v include/linux/mm_types.h" and it is showing
below error and warning which is linked to enum in discussion.

include/linux/mm_types.h:612: info: Scanning doc for typedef vm_fault_t
include/linux/mm_types.h:623: info: Scanning doc for enum
include/linux/mm_types.h:628: warning: contents before sections
include/linux/mm_types.h:660: error: Cannot parse enum!
1 errors
1 warnings

Shall I keep the documentation for enum or remove it from this patch ?

>
> > >
> > > > > + * This enum is used to track the VM_FAULT code return by page
> > > > > + * fault handlers.
> > > >
> > > >  * Page fault handlers return a bitmask of these values to tell the
> > > >  * core VM what happened when handling the fault.
> > > >
> > >
> > > --
> > > Sincerely yours,
> > > Mike.
> > >
> >
>
> --
> Sincerely yours,
> Mike.
>