All of lore.kernel.org
 help / color / mirror / Atom feed
From: tanxiaojun@huawei.com (Tan Xiaojun)
To: linux-arm-kernel@lists.infradead.org
Subject: [bug report & help] arm64: ltp testcase "migrate_pages01" failed
Date: Tue, 17 Oct 2017 10:58:53 +0800	[thread overview]
Message-ID: <59E571ED.9090106@huawei.com> (raw)
In-Reply-To: <59E49B2B.8020802@huawei.com>

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.
> 
> 
> .
> 

WARNING: multiple messages have this Message-ID (diff)
From: Tan Xiaojun <tanxiaojun@huawei.com>
To: <catalin.marinas@arm.com>, <will.deacon@arm.com>,
	<hanjun.guo@linaro.org>, <xieyisheng1@huawei.com>,
	<tglx@linutronix.de>, <punit.agrawal@arm.com>,
	<boris.ostrovsky@oracle.com>
Cc: <linux-kernel@vger.kernel.org>,
	"linux-arm-kernel@lists.infradead.org" 
	<linux-arm-kernel@lists.infradead.org>
Subject: Re: [bug report & help] arm64: ltp testcase "migrate_pages01" failed
Date: Tue, 17 Oct 2017 10:58:53 +0800	[thread overview]
Message-ID: <59E571ED.9090106@huawei.com> (raw)
In-Reply-To: <59E49B2B.8020802@huawei.com>

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.
> 
> 
> .
> 

  reply	other threads:[~2017-10-17  2:58 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-10-16 11:42 [bug report & help] arm64: ltp testcase "migrate_pages01" failed Tan Xiaojun
2017-10-16 11:42 ` Tan Xiaojun
2017-10-17  2:58 ` Tan Xiaojun [this message]
2017-10-17  2:58   ` Tan Xiaojun
2017-10-17  9:23   ` Will Deacon
2017-10-17  9:23     ` Will Deacon
2017-10-17 11:33     ` Tan Xiaojun
2017-10-17 11:33       ` Tan Xiaojun
2017-10-17 13:19     ` Yisheng Xie
2017-10-17 13:19       ` Yisheng Xie
2017-10-18  1:45       ` Yisheng Xie
2017-10-18  1:45         ` Yisheng Xie

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=59E571ED.9090106@huawei.com \
    --to=tanxiaojun@huawei.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.