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

[Alexander Viro]

On Sat, 29 Dec 2001, Andrew Morton wrote:

> > 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?)

For ->writepage() - nothing.  It _is_ asynchronous and it _can_ fail.
Due to failing allocations, IO errors, whatever.

Now, the fs consistency stuff is a different story.  Fixes had been
in -ac since before 2.4.0 and I distinctly remember at least one of
3 area getting synced with -linus.  My fault - I assumed that the
whole patch went there at that point.  I'll try to dig the rest out.

2.4.9-ac* is probably a good starting point - they are in generic_file_write()
and in __block_write_full_page()