linux-kernel - Re: [PATCH] mm/vmalloc: Check absolute error return from vmap_[p4d|pud|pmd|pte]

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <406afc57-5a77-a77c-7f71-df1e6837dae1@arm.com>
Date:   Thu, 13 Jun 2019 20:51:17 +0530
From:   Anshuman Khandual <anshuman.khandual@....com>
To:     Roman Penyaev <rpenyaev@...e.de>
Cc:     linux-mm@...ck.org, linux-kernel@...r.kernel.org,
        Rick Edgecombe <rick.p.edgecombe@...el.com>,
        Andrey Ryabinin <aryabinin@...tuozzo.com>,
        Mike Rapoport <rppt@...ux.ibm.com>,
        Roman Gushchin <guro@...com>, Michal Hocko <mhocko@...e.com>,
        "Uladzislau Rezki (Sony)" <urezki@...il.com>,
        Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [PATCH] mm/vmalloc: Check absolute error return from
 vmap_[p4d|pud|pmd|pte]_range()



On 06/13/2019 03:03 PM, Roman Penyaev wrote:
> On 2019-06-13 10:12, Anshuman Khandual wrote:
>> vmap_pte_range() returns an -EBUSY when it encounters a non-empty PTE. But
>> currently vmap_pmd_range() unifies both -EBUSY and -ENOMEM return code as
>> -ENOMEM and send it up the call chain which is wrong. Interestingly enough
>> vmap_page_range_noflush() tests for the absolute error return value from
>> vmap_p4d_range() but it does not help because -EBUSY has been merged with
>> -ENOMEM. So all it can return is -ENOMEM. Fix this by testing for absolute
>> error return from vmap_pmd_range() all the way up to vmap_p4d_range().
> 
> I could not find any real external caller of vmap API who really cares
> about the errno, and frankly why they should?  This is allocation path,

map_vm_area() which is an exported symbol suppose to provide the right
error code regardless whether it's current users care for it or not.

> allocation failed - game over.  When you step on -EBUSY case something
> has gone completely wrong in your kernel, you get a big warning in
> your dmesg and it is already does not matter what errno you get.

Its true that vmap_pte_range() does warn during error conditions. But if
we really dont care about error return code then we should just remove
specific error details (ENOMEM/EBUSY) and instead replace them with simple
boolean false/true or (0/1/-1) return values at each level. Will that be
acceptable ? What we have currently is wrong where vmap_pmd_range() could
just wrap EBUSY as ENOMEM and send up the call chain.