[CRIU] [PATCH] zdtm.py: check permissions for memory mappings
Andrew Vagin
avagin at virtuozzo.com
Fri Jul 15 11:47:21 PDT 2016
On Thu, Jul 14, 2016 at 02:02:55PM +0300, Pavel Emelyanov wrote:
> On 07/13/2016 08:19 PM, Andrew Vagin wrote:
> > On Wed, Jul 13, 2016 at 03:33:19PM +0300, Pavel Emelyanov wrote:
> >> On 06/27/2016 05:52 PM, Andrey Vagin wrote:
> >>> From: Andrew Vagin <avagin at virtuozzo.com>
> >>
> >> Andrey, will current criu-dev pass with this patch? If not what patches
> >> do we need to make it passing?
> >
> > It was attempt to wirte a test for Stas' patch:
> > proc_parse: fix vma file open mode recognition
> >
> > But it doesn't catch the origin problem and trigger another one:
> > https://github.com/xemul/criu/issues/180
>
> OK, so what do we do with __this__ patch:
>
> a) apply one and go on
> b) apply one and have a bunch of failures in jenkins
> c) drop one and wait for better version
We will fix the issue and commit this patch.
>
> ?
>
> >>
> >>> Signed-off-by: Andrew Vagin <avagin at virtuozzo.com>
> >>> ---
> >>> test/zdtm.py | 7 ++++---
> >>> 1 file changed, 4 insertions(+), 3 deletions(-)
> >>>
> >>> diff --git a/test/zdtm.py b/test/zdtm.py
> >>> index 9454399..7aed548 100755
> >>> --- a/test/zdtm.py
> >>> +++ b/test/zdtm.py
> >>> @@ -917,16 +917,17 @@ def get_visible_state(test):
> >>> for pid in pids:
> >>> files[pid] = set(os.listdir("/proc/%s/root/proc/%s/fd" % (test.getpid(), pid)))
> >>>
> >>> - cmaps = [[0, 0]]
> >>> + cmaps = [[0, 0, ""]]
> >>> last = 0
> >>> for mp in open("/proc/%s/root/proc/%s/maps" % (test.getpid(), pid)):
> >>> m = map(lambda x: int('0x' + x, 0), mp.split()[0].split('-'))
> >>> - if cmaps[last][1] == m[0]:
> >>> + m.append(mp.split()[1])
> >>> + if cmaps[last][1] == m[0] and cmaps[last][2] == m[2]:
> >>> cmaps[last][1] = m[1]
> >>> else:
> >>> cmaps.append(m)
> >>> last += 1
> >>> - maps[pid] = set(map(lambda x: '%x-%x' % (x[0], x[1]), cmaps))
> >>> + maps[pid] = set(map(lambda x: '%x-%x %s' % (x[0], x[1], x[2]), cmaps))
> >>>
> >>> cmounts = []
> >>> try:
> >>>
> >>
> > .
> >
>
More information about the CRIU
mailing list