2002-10-29 14:35:47

by Amol Kumar Lad

[permalink] [raw]
Subject: mlockall() with MCL_FUTURE

Hi,
I was just going through its implementation. If mlockall() is invoked
with MCL_FUTURE, does it mean that all the existing locked mappings of
process should get unlocked ? Attaching code segment from do_mlockall().
I am using 2.4.18 kernel

static int do_mlockall(int flags)
{
...
newflags = vma->vm_flags | VM_LOCKED;
if (!(flags & MCL_CURRENT))
newflags &= ~VM_LOCKED;
error = mlock_fixup(vma, vma->vm_start, vma->vm_end,
newflags);
...
}

please cc me.

thanks
Amol




2002-10-29 14:55:24

by Tom Rini

[permalink] [raw]
Subject: Re: mlockall() with MCL_FUTURE

On Tue, Oct 29, 2002 at 08:11:46PM -0500, Amol Kumar Lad wrote:
> Hi,
> I was just going through its implementation. If mlockall() is invoked
> with MCL_FUTURE, does it mean that all the existing locked mappings of
> process should get unlocked ? Attaching code segment from do_mlockall().
> I am using 2.4.18 kernel

There is a problem here which is fixed in 2.4.19, I believe.

--
Tom Rini (TR1265)
http://gate.crashing.org/~trini/