|
From: Tom H. <to...@co...> - 2005-11-02 14:52:08
|
Does the as_mmap test serve any purpose now? It tries to map a chunk of memory at various fixed addresses in memory - presumably the idea was to check that it couldn't map in the valgrind areas. Certainly both sets of expected results show it failing at the high addresses. With the address space manager there are no longer any real guarantees about which addresses might or might not be available and one of the areas that is expected to fail is actually working on amd64. Tom -- Tom Hughes (to...@co...) http://www.compton.nu/ |
|
From: Tom H. <to...@co...> - 2005-11-02 14:57:20
|
In message <yek...@de...>
Tom Hughes <to...@co...> wrote:
> Does the as_mmap test serve any purpose now?
>
> It tries to map a chunk of memory at various fixed addresses in
> memory - presumably the idea was to check that it couldn't map in
> the valgrind areas. Certainly both sets of expected results show
> it failing at the high addresses.
>
> With the address space manager there are no longer any real guarantees
> about which addresses might or might not be available and one of the
> areas that is expected to fail is actually working on amd64.
Much the same applies to as_shm as well, which is trying to attach a
shared memory segment at a high address and expecting it to fail.
Tom
--
Tom Hughes (to...@co...)
http://www.compton.nu/
|
|
From: Nicholas N. <nj...@cs...> - 2005-11-02 15:26:03
|
On Wed, 2 Nov 2005, Tom Hughes wrote: >> Does the as_mmap test serve any purpose now? >> >> It tries to map a chunk of memory at various fixed addresses in >> memory - presumably the idea was to check that it couldn't map in >> the valgrind areas. Certainly both sets of expected results show >> it failing at the high addresses. >> >> With the address space manager there are no longer any real guarantees >> about which addresses might or might not be available and one of the >> areas that is expected to fail is actually working on amd64. > > Much the same applies to as_shm as well, which is trying to attach a > shared memory segment at a high address and expecting it to fail. Removing both sounds fine to me. Good job with cleaning up the regtests, Tom. A bad test is arguably worse than no test at all. Nick |
|
From: Julian S. <js...@ac...> - 2005-11-02 15:32:58
|
> > Much the same applies to as_shm as well, which is trying to attach a > > shared memory segment at a high address and expecting it to fail. > > Removing both sounds fine to me. Me too. mremap2 is another problem case. I wrote that to test the mremap handling really carefully, but it does assume that kernel and V make the same layout choices when doing mremap. So it's not robust. Problem is I can't think of a way make it robust without weakening the testing it does. J |