Re: [LKP] [mm] Find a way to test remap_file_pages

[Vlastimil Babka <vbabka@suse.cz>]

On 01/12/2015 12:19 PM, Kirill A. Shutemov wrote:
> On Mon, Jan 12, 2015 at 02:26:29PM +1100, Stephen Rothwell wrote:
>> [Just cc'ing Andrew as this patch is in his tree]
>> 
>> On Mon, 12 Jan 2015 11:19:01 +0800 Huang Ying <ying.huang@intel.com> wrote:
>> >
>> > FYI, we noticed the below changes on
>> > 
>> > git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
>> > commit 974bb1d5d44f008d7f207437d63d46658ff41dff ("mm: replace remap_file_pages() syscall with emulation")
>> > 
>> > That is, trinity can be used to test some aspect of remap_file_pages.

Well, yeah :) I think the number of bugs trinity has already found in this area
during the last year, was one of the reasons behind its deprecation and
eventually removal.

>> > 
>> > +---------------------------------------------------+------------+------------+
>> > |                                                   | d96eeaf2cb | 974bb1d5d4 |
>> > +---------------------------------------------------+------------+------------+
>> > | boot_successes                                    | 0          | 0          |
>> > | boot_failures                                     | 10         | 10         |
>> > | WARNING:at_net/netlink/genetlink.c:#genl_unbind() | 10         | 5          |
>> > | backtrace:netlink_setsockopt                      | 3          | 2          |
>> > | backtrace:SyS_setsockopt                          | 3          | 2          |
>> > | backtrace:SyS_socketcall                          | 3          | 2          |
>> > | backtrace:do_group_exit                           | 10         | 5          |
>> > | backtrace:SyS_exit_group                          | 10         | 5          |
>> > | page_allocation_failure:order:#,mode              | 0          | 5          |
>> > | backtrace:btrfs_test_extent_io                    | 0          | 5          |
>> > | backtrace:init_btrfs_fs                           | 0          | 5          |
>> > | backtrace:kernel_init_freeable                    | 0          | 5          |
>> > +---------------------------------------------------+------------+------------+
>> > 
>> > 
>> > [   29.665081] Bits 55-60 of /proc/PID/pagemap entries are about to stop being page-shift some time soon. See the linux/Documentation/vm/pagemap.txt for details.
>> > [   29.675960] mmap: trinity-c0 (873) uses deprecated remap_file_pages() syscall. See Documentation/vm/remap_file_pages.txt.

Is the warning above even related to the warning below, or just coincidence? The
remap_file_pages() deprecation warning doesn't generate stacktrace AFAICS.

>> > [   29.681044] ------------[ cut here ]------------
>> > [   29.681841] WARNING: CPU: 0 PID: 870 at net/netlink/genetlink.c:1037 genl_unbind+0x96/0xaa()
>> > [   29.683610] warning: process `trinity-c0' used the obsolete bdflush system call
>> > [   29.684911] Fix your initscripts?
>> > [   29.685683] VFS: Warning: trinity-c0 using old stat() call. Recompile your binary.
>> > [   29.686963] CPU: 0 PID: 870 Comm: trinity-main Not tainted 3.19.0-rc3-next-20150109-g06f367d #290
>> > [   29.688399] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
>> > [   29.689349]  00000000 00000000 d34f3ea4
>> > [   29.694083] ------------[ cut here ]------------
>> > [   29.695096] WARNING: CPU: 0 PID: 872 at net/netlink/genetlink.c:1037 genl_unbind+0x96/0xaa()
>> > [   29.696724]  c15933f7 d34f3ec0 c103c7ff 0000040d c1581310
>> > [   29.698051]  00000001 00000001 00000000 d34f3ed0 c103c89c 00000009 00000000 d34f3ee8
>> > [   29.699904] CPU: 0 PID: 872 Comm: trinity-main Not tainted 3.19.0-rc3-next-20150109-g06f367d #290
>> > [   29.701406] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
>> > [   29.702291]  00000000 00000000 d34d9ea4 c15933f7 d34d9ec0
>> > [   29.703253]  c1581310 c18836fc 00000001 d341f840 00000000 d34f3f0c c157f810 d15b0b40
>> > [   29.705361] Call Trace:
>> > [   29.705868]  [<c15933f7>] dump_stack+0x16/0x18
>> > [   29.706652]  [<c103c7ff>] warn_slowpath_common+0x79/0x90
>> > [   29.707496]  [<c1581310>] ? genl_unbind+0x96/0xaa
>> > [   29.708295]  [<c103c89c>] warn_slowpath_null+0xf/0x13
>> > [   29.709142]  [<c1581310>] genl_unbind+0x96/0xaa
>> > [   29.709943]  [<c157f810>] netlink_release+0x2d7/0x2ed
>> > [   29.710776]  [<c1553efd>] sock_release+0x10/0x5a
>> > [   29.711531]  [<c1553f52>] sock_close+0xb/0xf
>> > [   29.712286]  [<c1109a69>] __fput+0xf7/0x1f0
>> > [   29.713044]  [<c1109b88>] ____fput+0x8/0xa
>> > [   29.713794]  [<c1052412>] task_work_run+0x4e/0x70
>> > [   29.714566]  [<c103e771>] do_exit+0x43a/0x9c7
>> > [   29.715300]  [<c103ed6f>] do_group_exit+0x4a/0xd6
>> > [   29.716111]  [<c103ee0c>] SyS_exit_group+0x11/0x11
>> > [   29.716975]  [<c1598cfd>] syscall_call+0x7/0x7
>> > [   29.717722]  [<c1590000>] ? io_submit_one+0x5b3/0x72a
>> > [   29.719207]  c103c7ff 0000040d c1581310
>> > [   29.720122]  00000001 00000001 00000000 d34d9ed0 c103c89c 00000009 00000000 d34d9ee8
>> > [   29.721973]  c1581310 c18836fc 00000004 d0ebb840 00000000 d34d9f0c c157f810
>> > [   29.723252] ---[ end trace 28426633f87b780a ]---
>> > [   29.724282] ------------[ cut here ]------------
> 
> 
> It doesn't make much sense to me. I don't see anything
> remap_file_pages()-related here. Or even VMA related.
> 
> And why now? The commit is in Andrew's tree for few release cycles.
> 
> Could you elaborate on what you observe? And what the table above means?
>