Re: [RFC PATCH] PM / hibernate: make sure each resuming page is in current memory zones

[chenyu5 <yu.c.chen@intel.com>]

Hi, Joey
thanks for your reply,

On 2015年07月02日 14:19, joeyli wrote:
> Hi Chen Yu,
>
> On Tue, Jun 30, 2015 at 11:54:28PM +0800, Chen Yu wrote:
>> Commit 84c91b7ae07c ("PM / hibernate: avoid unsafe pages in e820 reserved
>> regions") was reverted, because this patch makes resume from hibernation
>> on Lenovo x230 unreliable. But reverting may bring back the kernel
>> exception firstly reported in former patch. In general, there are three
>> problems in current code when resuming from hibernation:
>>
>> 1.Resuming page may also be in second kernel's e820 reserved region.
>> BIOS-e820: [mem 0x0000000069d4f000-0x0000000069e12fff] reserved
>> this causes kernel exception described in Commit 84c91b7ae07c
>> ("PM / hibernate: avoid unsafe pages in e820 reserved regions")
>>
>> 2.If Commit 84c91b7ae07c ("PM / hibernate: avoid unsafe pages in e820
>> reserved regions") is applied to fix problem 1, and if E820_RESERVED_KERN
>> regions causes some regions at e820 table not page aligned,
>> e820_mark_nosave_regions will misjudgment the non-page aligned space to
>> be "hole" space and add to nosave regions, this causes resuming failed.
>> Refer to https://bugzilla.kernel.org/show_bug.cgi?id=96111 for detail.
>>
>
> Sorry for I am not fully understand why checking pfn_valid could avoid
> the above 2. issue?
>
[Yu] According to e820_mark_nosave_regions, two kinds of e820
regions will be regarded as nosave:
a.e820.map holes between each e820entry
b.e820entry with (!E820_RAM && !E820_RESERVED_KERN)
dmesg in
https://bugzilla.kernel.org/show_bug.cgi?id=96111
shows that:

BIOS-e820: [mem 0x000000005baff000-0x00000000d684ffff] usable
reserve setup_data: [mem 0x000000009d3e0018-0x000000009d3f0057] usable
PM: Registered nosave memory: [mem 0x9d3e0000-0x9d3e0fff]

e820 map was changed by situation a, but it should not be added to
nosave list, right?

and according to saveable_page, only pages in the zone will be saved:
if (page_zone(page_ != zone)
	return NULL;

The function of is_valid_orig_page in this patch treats setup_data as
valid page, so I think this workaround issue 2.


> Per my understood should waiting Yinghai Lu's patches for killing
> E820_RESERVED_KERN to avoid the aligned problem.
[Yu] yes, it is a graceful solution, but I don't see that in 4.1? And
we encountered problem 3 in our testing enviroment, so I wrote a patch
to deal with it first.
>
>
> Thanks a lot!
> Joey Lee
>

Best Regards,
Yu