On Fri, Jul 12, 2013 at 03:04:41PM +0100, Andrew Cooper wrote:
> On 12/07/13 14:48, Daniel Kiper wrote:
> > get_pt_note_info() always ignores VMCOREINFO_XEN note
> > because buf size is too small. It does not have place
> > for \0 char which marks EOS. This patch fixes that bug
> > and VMCOREINFO_XEN note living in /proc/vmcore file
> > could be properly detected now.
> >
> > Signed-off-by: Daniel Kiper <daniel.kiper at oracle.com>
> > ---
> > elf_info.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/elf_info.c b/elf_info.c
> > index 0c1e36a..70a6dd2 100644
> > --- a/elf_info.c
> > +++ b/elf_info.c
> > @@ -310,7 +310,7 @@ get_pt_note_info(void)
> > {
> > int n_type, size_name, size_desc;
> > off_t offset, offset_desc;
> > - char buf[VMCOREINFO_XEN_NOTE_NAME_BYTES];
> > + char buf[VMCOREINFO_XEN_NOTE_NAME_BYTES + 1];
> > char note[MAX_SIZE_NHDR];
> >
> > nr_cpus = 0;
>
> Elf Note namesz is defined to include the terminating null character, so
> I would argue that VMCOREINFO_XEN_NOTE_NAME_BYTES is off-by-one.
I have checked that once again. By mistake I have assumed that
sizeof("string") does not take into account \0 char (like strlen()).
It means that in real buf could accommodate VMCOREINFO_XEN string
with EOS and relevant condition should be fixed instead of buf size.
I will fix it in second patch series release.
Thanks,
Daniel
