Re: [PATCH] error path in setup_arg_pages() misses vm_unacct_memory()

[Hugh Dickins <hugh@veritas.com>]

On Wed, 14 Sep 2005, Kirill Korotaev wrote:
> 
> is vma accounting in arch/x86_64/ia32/syscall32.c and arch/ppc64/kernel/vdso.c
> removed due to fixed size of vma (vsyscall/vdso mappings)?
> in other respects it looks ok.

It's removed because without VM_ACCOUNT it was steadily leaking:
the main path was wrong, never mind the error path.  We could add
VM_ACCOUNT to the flags to fix that, but in 99.999% of cases we
should not be accounting these - all mms are sharing the same page.

Ben designed it to allow for gdb setting breakpoints in the vDSO
page (COW then giving a private page), but that's an very unusual
case, which isn't handled quite right anyway: we need to take a
separate look at that sometime - we account for what's VM_WRITE,
not for what ptrace might write to by the backdoor.

> > So x86_64 32-bit and ppc64 vDSO ELFs have been leaking memory into
> > Committed_AS each time they're run.  But don't add VM_ACCOUNT to them,
> > it's inappropriate to reserve against the very unlikely case that gdb
> > be used to COW a vDSO page - we ought to do something about that in
> > do_wp_page, but there are yet other inconsistencies to be resolved.

Hugh