Re: ramdisk corruption problems - was: RE: pivot_root and initrd kern

[Andrew Morton]

Alexander Viro wrote:
> 
> On Sat, 29 Dec 2001, Andrew Morton wrote:
> 
> > Would it be necessary to preallocate the holes at mmap() time?  Mad
> > hand-waving: Could we not perform the instantiation at pagefault time,
> > and give the caller SIGBUS if we cannot allocate the blocks?  Or if
> > there's an IO error, or quota exceeded.
> 
> Allocation at mmap() Is Not Going To Happen.  Consider it vetoed.
> There are applications that use mmap() on large and very sparse
> files.

I think Andrea was referring to simply reserving the necessary
amount of disk space, rather than actually instantiating the
blocks.  But even that would be a big problem for the applications
which you describe.

> > Question: can someone please define BH_New?  Its lifecycle seems
> > very vague.  We never actually seem to *clear* it anywhere for
> > ext2, and it appears that the kernel will keep on treating a
> > clearly non-new buffer as "new" all the time.  ext3 explicitly
> > clears BH_New in get_block(), if it finds the block was already
> > present in the file.  I did this because we need the newness
> > info for internal purposes.
> 
> It should be reset when we submit IO.

well...  It isn't.  And I'd like a chance to review/test any
proposed changes in this area which are outside specific filesystems...

> Breakage related to failing allocation is indeed not new, but
> that's a long story.  And no, "allocate on mmap()" is not a fix.

Yup.  But what *is* the fix?  (filemap_nopage?)

-