linux-kernel - Re: [bug report & help] arm64: ltp testcase "migrate

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]

Date:   Tue, 17 Oct 2017 10:58:53 +0800
From:   Tan Xiaojun <tanxiaojun@...wei.com>
To:     <catalin.marinas@....com>, <will.deacon@....com>,
        <hanjun.guo@...aro.org>, <xieyisheng1@...wei.com>,
        <tglx@...utronix.de>, <punit.agrawal@....com>,
        <boris.ostrovsky@...cle.com>
CC:     <linux-kernel@...r.kernel.org>,
        "linux-arm-kernel@...ts.infradead.org" 
        <linux-arm-kernel@...ts.infradead.org>
Subject: Re: [bug report & help] arm64: ltp testcase "migrate_pages01" failed

Hi all,

I'm not sure if this is the problem on arm64 numa. What do you think ?
By the way, this testcase can be successful in any case on x86.

Thanks.
Xiaojun.

On 2017/10/16 19:42, Tan Xiaojun wrote:
> Hi all,
> 
> I test ltp in Hisilicon D05 board and get a failed result about the testcase "migrate_pages01".
> 
> In fact, The sub testcase "test_invalid_nodes" failed. The testcase is to find a invalid numa node and migrate memory pages to this node via syscall of "migrate_pages".
> The expected result of this case is returning "-1", but it actually return "0".
> 
> --------------------------------------------------------
> # ./migrate_pages01
> migrate_pages01    0  TINFO  :  test_empty_mask
> migrate_pages01    1  TPASS  :  expected ret success: returned value = 0
> migrate_pages01    0  TINFO  :  test_invalid_pid -1
> migrate_pages01    2  TPASS  :  expected ret success: returned value = -1
> migrate_pages01    3  TPASS  :  expected failure: TEST_ERRNO=ESRCH(3): No such process
> migrate_pages01    0  TINFO  :  test_invalid_pid unused pid
> migrate_pages01    4  TPASS  :  expected ret success: returned value = -1
> migrate_pages01    5  TPASS  :  expected failure: TEST_ERRNO=ESRCH(3): No such process
> migrate_pages01    0  TINFO  :  test_invalid_masksize
> migrate_pages01    6  TPASS  :  expected ret success: returned value = -1
> migrate_pages01    7  TPASS  :  expected failure: TEST_ERRNO=EINVAL(22): Invalid argument
> migrate_pages01    0  TINFO  :  test_invalid_mem -1
> migrate_pages01    8  TPASS  :  expected ret success: returned value = -1
> migrate_pages01    9  TPASS  :  expected failure: TEST_ERRNO=EFAULT(14): Bad address
> migrate_pages01    0  TINFO  :  test_invalid_mem invalid prot
> migrate_pages01   10  TPASS  :  expected ret success: returned value = -1
> migrate_pages01   11  TPASS  :  expected failure: TEST_ERRNO=EFAULT(14): Bad address
> migrate_pages01    0  TINFO  :  test_invalid_mem unmmaped
> migrate_pages01   12  TPASS  :  expected ret success: returned value = -1
> migrate_pages01   13  TPASS  :  expected failure: TEST_ERRNO=EFAULT(14): Bad address
> migrate_pages01    0  TINFO  :  test_invalid_nodes
> migrate_pages01   14  TFAIL  :  migrate_pages_common.c:45: unexpected failure - returned value = 0, expected: -1
> migrate_pages01   15  TFAIL  :  migrate_pages_common.c:55: call succeeded unexpectedly
> migrate_pages01    0  TINFO  :  test_invalid_perm
> migrate_pages01   16  TPASS  :  expected ret success: returned value = -1
> migrate_pages01   17  TPASS  :  expected failure: TEST_ERRNO=EPERM(1): Operation not permitted
> --------------------------------------------------------
> 
> I debug and find a interesting thing, this case does not always fail.
> 
> 1) If one or several numa nodes have no memory, this case will run successfully like below:
> 
> --------------------
> available: 4 nodes (0-3)
> node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
> node 0 size: 65309 MB
> node 0 free: 61650 MB
> node 1 cpus: 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
> node 1 size: 65404 MB
> node 1 free: 61377 MB
> node 2 cpus: 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
> node 2 size: 65401 MB
> node 2 free: 62316 MB
> node 3 cpus: 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63
> node 3 size: 0 MB
> node 3 free: 0 MB
> node distances:
> node   0   1   2   3
>   0:  10  15  20  20
>   1:  15  10  20  20
>   2:  20  20  10  15
>   3:  20  20  15  10
> ---------------------
> 
> This testcase will find node number 3 and migrate pages to node 3. And syscall of "migrate_pages" return -1, test succeeded.
> 
> 2) In most cases, all nodes have memory, and the testcase will get non-existent node like node number 4. The syscall of "migrate_pages" should also return -1, but return 0 actually.
> So the testcase failed.
> 
> I think it is a problem in arm64. But I am not familiar with numa, so I ask for help from you.
> 
> Thanks.
> Xiaojun.
> 
> 
> .
>