Re: [PATCH for-4.19?] xen/vmap: Document the vmap header

["Alejandro Vallejo" <alejandro.vallejo@xxxxxxxxx>]

On Wed Jul 3, 2024 at 8:41 AM BST, Jan Beulich wrote:
> On 02.07.2024 15:03, Alejandro Vallejo wrote:
> > --- a/xen/include/xen/vmap.h
> > +++ b/xen/include/xen/vmap.h
> > @@ -1,34 +1,131 @@
> > +/*
> > + * Interface to map physical memory onto contiguous virtual memory areas.
> > + *
> > + * Two ranges of linear address space are reserved for this purpose: A 
> > general
> > + * purpose area (VMAP_DEFAULT) and a livepatch-specific area (VMAP_XEN). 
> > The
> > + * latter is used when loading livepatches and the former for everything 
> > else.
> > + */
> >  #if !defined(__XEN_VMAP_H__) && defined(VMAP_VIRT_START)
> >  #define __XEN_VMAP_H__
> >  
> >  #include <xen/mm-frame.h>
> >  #include <xen/page-size.h>
> >  
> > +/** Identifiers for the linear ranges tracked by vmap */
>
> Here and below: Why /** ? Seeing ...
>
> >  enum vmap_region {
> > +    /*
> > +     * Region used for general purpose RW mappings. Mapping/allocating 
> > memory
> > +     * here can induce extra allocations for the supporting page tables.
> > +     */
>
> ... this vs ...
>
> >      VMAP_DEFAULT,
> > +    /**
> > +     * Region used for loading livepatches. Can't use VMAP_DEFAULT because 
> > it
> > +     * must live close to the running Xen image. The caller also ensures 
> > all
> > +     * page tables are already in place with adequate PTE flags.
> > +     */
>
> ... this, it's not even looking to be consistent.
>
> >      VMAP_XEN,
> > +    /** Sentinel value for bounds checking */
> >      VMAP_REGION_NR,
> >  };

Double stars are Doxygen style for documenting interfaces. I don't mind getting
rid of those. Some functions exhibit them in the arm port, but it's true
virtually nothing in common Xen does. VMAP_XEN's comment was meant to have it
too, I just missed it :)

> >  
> > +/**
> > + * Runtime initialiser for each vmap region type
> > + *
> > + * Must only be called once per vmap region type.
> > + *
> > + * @param type  Designation of the region to initialise.
> > + * @param start Start address of the `type` region.
> > + * @param end   End address (not inclusive) of the `type` region
> > + */
> >  void vm_init_type(enum vmap_region type, void *start, void *end);
> >  
> > +/**
> > + * Maps a range of physical ranges onto a single virtual range
>
> Largely related to it not really being clear what "a range of physical ranges"
> actually is, maybe "a set of physical ranges"?

Ack. I like your phrasing better.

>
> > + * `mfn` is an array of `nr` physical ranges, each of which is 
> > `granularity`
> > + * pages wide. `type` defines which vmap region to use for the mapping and
> > + * `flags` is the PTE flags the page table leaves are meant to have.
> > + *
> > + * Typically used via the vmap() and vmap_contig() helpers.
> > + *
> > + * @param mfn          Array of mfns
> > + * @param granularity  Number of contiguous pages each mfn represents
> > + * @param nr           Number of mfns in the `mfn` array
> > + * @param align        Alignment of the virtual area to map
> > + * @param flags        PTE flags for the leaves of the PT tree.
> > + * @param type         Which region to create the mappings on
> > + */
> >  void *__vmap(const mfn_t *mfn, unsigned int granularity, unsigned int nr,
> >               unsigned int align, unsigned int flags, enum vmap_region 
> > type);
> > +
> > +/**
> > + * Map an arrray of pages contiguously into the VMAP_DEFAULT vmap region
>
> Nit: One r too many in "array".

Sure.

>
> > + * @param[in] mfn Pointer to the base of an array of mfns
> > + * @param[in] nr  Number of mfns in the array
> > + */
> >  void *vmap(const mfn_t *mfn, unsigned int nr);
> > +
> > +/**
> > + * Maps physically contiguous pages onto the VMAP_DEFAULT vmap region
> > + *
> > + * Used when the directmap is unavailable (i.e: due to secret hiding)
>
> Please omit this. There's no reason to suggest it can't also be used for
> other purposes.

Sure.

>
> > + * @param mfn Base mfn of the physical region
> > + * @param nr  Number of mfns in the physical region
> > + */
> >  void *vmap_contig(mfn_t mfn, unsigned int nr);
> > +
> > +/**
> > + * Unmaps a range of virtually contiguous memory from one of the vmap 
> > regions
> > + *
> > + * The system remembers internally how wide the mapping is and unmaps it 
> > all.
> > + * It also can determine the vmap region type from the `va`.
> > + *
> > + * @param va Virtual base address of the range to unmap
> > + */
> >  void vunmap(const void *va);
> >  
> > +/**
> > + * Allocate `size` octets of possibly non-contiguous physical memory and 
> > map
> > + * them contiguously in the VMAP_DEFAULT vmap region
> > + *
> > + * The system remembers internally how wide the mapping is and unmaps it 
> > all.
>
> This is a verbatim copy of what you say for vunmap(), yet it looks unrelated
> here. Maybe you meant to edit it after copy-and-paste? Or maybe this was
> meant to go ...

Yeah, that shouldn't be there and was meant to go...

>
> > + * @param size Pointer to the base of an array of mfns
> > + * @return Pointer to the mapped area on success; NULL otherwise.
> > + */
> >  void *vmalloc(size_t size);
> > +
> > +/** Same as vmalloc(), but for the VMAP_XEN vmap region. */
> >  void *vmalloc_xen(size_t size);
> >  
> > +/** Same as vmalloc(), but set the contents to zero before returning */
> >  void *vzalloc(size_t size);
> > +
> > +/**
> > + * Unmap and free memory from vmalloc(), vmalloc_xen() or vzalloc()
> > + *
> > + * @param va Virtual base address of the range to free and unmap
> > + */
> >  void vfree(void *va);
>
> ... here?

... there.

>
> Also, if you want this to be considered for 4.19, please don't forget to Cc
> Oleksii.
>
> Jan

Will do on v2.

Cheers,
Alejandro