Re: truncate on MAP_SHARED files in ramfs filesystems on no-mmu

[Mike Frysinger <vapier.adi@gmail.com>]

On Thu, Jul 9, 2009 at 06:06, David Howells wrote:
> Mike Frysinger wrote:
>> reviewing LTP tests shows mmap09 failing.  this test creates a file of
>> a certain length, opens it and creates a shared mapping, and then
>> tries various truncate tests:
>> truncate to a smaller size
>> truncate to a larger size
>> truncate to size 0
>>
>> the first and last fail on no-mmu due to
>> file-nommu.c:ramfs_nommu_resize() rejecting attempts to shrink on a
>> shared mapping:
>
> Yes.  That's exactly right.
>
>> my question is why?  if an application maps a fd with MAP_SHARED,
>> truncates it, and then it or someone else who has that fd mapped tries
>> to access the now-invalid tail end, that is a bug in the application.
>> i dont see why we should be protecting users here from their own buggy
>> code ?
>
> This is protecting the kernel as much as the user.  There's no MMU to enforce
> denial of access on the pages that truncate returns to the system.

you dont need a MMU (virtual memory) to protect against it.  you only
need a MPU which some systems have.

> This doesn't only protect the process with a mapping on that file against its
> own truncate, but also other processes that have made mappings against that
> file.

and those too are broken

> Whilst you can argue it either way, you need a better reason to change this
> than it causes some LTP failures.  You cannot expect all the MM-related LTP
> tests to work against a NOMMU system.

crappy programming is likely to crash regardless of standard functions
we attempt to disable in the kernel.  this isnt a virtual memory issue
at all, it's memory protection.

> Doing it this way also makes things simpler in the kernel and makes the system
> more robust.

really ?  looks like the kernel is a lot more complicated to me.  the
fix here would be to delete a whole bunch of code.

> If a process shared mmaps a file and then wants to truncate it, it can always
> munmap the excess first.

sure, we could go around changing a whole bunch of things specific to
no-mmu, but that's kind of the wrong way to go.  applications shouldnt
need to know they're running with different MMU features available.
-mike