Hi Steve, It looks like the first patch I sent (above) was not the latest version. I found it when testing one other cifsacl patch today. Attached the fix for this bug. Also attached xfstest for verifying that it works. @Aurélien Aptel could you please review the test and add it to the buildbot for cifsacl and modefromsid? When none of those two are used, the test will fail. Regards, Shyam On Mon, Nov 9, 2020 at 9:46 AM Shyam Prasad N <nspmangalore@xxxxxxxxx> wrote: > > Also attaching the sticky bit implementation. > Have tested out some positive and negative test cases too. Works well, > from what I can tell. > > Regards, > Shyam > > On Mon, Nov 9, 2020 at 11:12 PM Shyam Prasad N <nspmangalore@xxxxxxxxx> wrote: > > > > Hi Steve, > > > > Here's the patch as per the discussion. I'm trying to maintain a > > preferred order of ACEs as much as possible. I had to modify the > > reverse conversion logic, since deny ACEs can appear in the middle of > > the list now. > > > > Tested thoroughly with many permission modes, conversions and reverse > > conversions. > > > > FYI, there's a sticky bit implementation which I've patched on top of > > this fix. Will send that for review soon. > > > > -- > > -Shyam > > > > -- > -Shyam -- Regards, Shyam
Attachment:
0001-cifs-Fix-cifsacl-ACE-mask-for-group-and-others.patch
Description: Binary data
Attachment:
002-modecheck.out
Description: Binary data
Attachment:
002-modecheck
Description: Binary data
