When we do restore VMA area after checkpoint we would like
to know if the area was locked or say it had mergeable attribute,
but at moment the kernel does not provide such information, thus
we can't figure out if we should call mlock/madvise on VMA restore.
This patch adds new VmFlags field to sma
On Fri, Oct 19, 2012 at 01:28:35AM +0400, Cyrill Gorcunov wrote:
> >
> > A common way in which we do this future-proofing is to display the info
> > in name:value tuples (eg, /proc/meminfo). So userspace parses for the
> > "name" rather than looking into a fixed position in the /proc output.
> >
On Thu, Oct 18, 2012 at 02:02:59PM -0700, Andrew Morton wrote:
> > +#ifdef CONFIG_CHECKPOINT_RESTORE
> > +static void show_smap_vma_flags(struct seq_file *m, struct vm_area_struct
> > *vma)
> > +{
> > + seq_printf(m,
> > + "VmFlags: %c%c%c%c%c%c%c%c%c\n",
> > +
If yes, I would like to
> > know if the output format provided below looks reasonable.
> >
> > Please review, thanks!
> > ---
>
> Also I've had a bit different output format, not sure
> which one is better.
> ---
> From: Cyrill Gorcunov
> Subj
s reasonable.
>
> Please review, thanks!
> ---
Also I've had a bit different output format, not sure
which one is better.
---
From: Cyrill Gorcunov
Subject: [RFC] procfs: Add VmFlags field in smaps output
When we do restore VMA area after checkpoint
we would like to know if the are
VmFlags field in smaps output
When we do restore VMA area after checkpoint
we would like to know if the area was locked
or say it has mergeable attribute, but at moment
the kernel does not provide such information, thus
we can't figure out if we should call mlock/madvise
on VMA restore.
6 matches
Mail list logo