Re: [PATCH] 2.5.21 "I can't get no compilation"

[Anton Altaparmakov]

At 12:54 10/06/02, Martin Dalecki wrote:
>Anton Altaparmakov wrote:
>>At 12:19 10/06/02, Martin Dalecki wrote:
>>
>>>The subject says it all...
>>>
>>>Contrary to other proposed patches I realized that there is
>>>no such thing as vmalloc_dma.
>>
>>Perhaps you ought to look in mm/vmalloc.c which contains:
>>void * vmalloc_dma (unsigned long size)
>>{
>>         return __vmalloc(size, GFP_KERNEL|GFP_DMA, PAGE_KERNEL);
>>}
>>Or are you going to tell me that is a figment of my imagination?
>
>Oh I have missed the chunk which delets it there apparently, since
>*nobody* is using this.

It could be used by out of kernel tree code. (Note I don't know any code 
that does...)

>The only place where a special
>__vmalloc setup code is used in nfs which GFP_NOFS flag added,
>but not the above. so providing vmalloc_nofs would make more
>sense then vmalloc_dma.

NTFS defines its own vmalloc_nofs (fs/ntfs/malloc.h) so if you intend to 
add a generic vmalloc_nofs, please remove the ntfs one (or ntfs will break)...

btw. the ntfs definition is:

static inline void *vmalloc_nofs(unsigned long size)
{
         if (likely(size >> PAGE_SHIFT < num_physpages)
                 return __vmalloc(size, GFP_NOFS | __GFP_HIGHMEM, PAGE_KERNEL);
         return NULL;
}

Best regards,

         Anton