lkml.org 
[lkml]   [2019]   [Dec]   [10]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] mm/hotplug: Only respect mem= parameter during boot stage
On 12/09/19 at 12:11pm, David Hildenbrand wrote:
> On 09.12.19 12:08, Jürgen Groß wrote:
> > On 09.12.19 12:01, David Hildenbrand wrote:
> >> On 09.12.19 11:24, Jürgen Groß wrote:
> >>> On 09.12.19 11:07, Michal Hocko wrote:
> >>>> On Fri 06-12-19 23:05:24, Baoquan He wrote:
> >>>>> In commit 357b4da50a62 ("x86: respect memory size limiting via mem=
> >>>>> parameter") a global varialbe global max_mem_size is added to store
> >>>>> the value which is parsed from 'mem= '. This truly stops those
> >>>>> DIMM from being added into system memory during boot.
> >>>>>
> >>>>> However, it also limits the later memory hotplug functionality. Any
> >>>>> memory board can't be hot added any more if its region is beyond the
> >>>>> max_mem_size. System will print error like below:
> >>>>>
> >>>>> [ 216.387164] acpi PNP0C80:02: add_memory failed
> >>>>> [ 216.389301] acpi PNP0C80:02: acpi_memory_enable_device() error
> >>>>> [ 216.392187] acpi PNP0C80:02: Enumeration failure
> >>>>>
> >>>>> >From document of 'mem =' parameter, it should be a restriction during
> >>>>> boot, but not impact the system memory adding/removing after booting.
> >>>>>
> >>>>> mem=nn[KMG] [KNL,BOOT] Force usage of a specific amount of memory
> >>>>>
> >>>>> So fix it by also checking if it's during SYSTEM_BOOTING stage when
> >>>>> restrict memory adding. Otherwise, skip the restriction.
> >>>>
> >>>> Could you be more specific about why the boot vs. later hotplug makes
> >>>> any difference? The documentation is explicit about the boot time but
> >>>> considering this seems to be like that since ever I strongly suspect
> >>>> that this is just an omission.
> >>>>
> >>>> Btw. how have you tested the situation fixed by 357b4da50a62?
> >>>
> >>> I guess he hasn't.
> >>>
> >>> The backtrace of the problem at that time was:
> >>>
> >>> [ 8321.876844] [<ffffffff81019ab9>] dump_trace+0x59/0x340
> >>> [ 8321.882683] [<ffffffff81019e8a>] show_stack_log_lvl+0xea/0x170
> >>> [ 8321.889298] [<ffffffff8101ac31>] show_stack+0x21/0x40
> >>> [ 8321.895043] [<ffffffff81319530>] dump_stack+0x5c/0x7c
> >>> [ 8321.900779] [<ffffffff8107fbf1>] warn_slowpath_common+0x81/0xb0
> >>> [ 8321.907482] [<ffffffff81009f54>] xen_alloc_pte+0x1d4/0x390
> >>> [ 8321.913718] [<ffffffff81064950>]
> >>> pmd_populate_kernel.constprop.6+0x40/0x80
> >>> [ 8321.921498] [<ffffffff815ef0a8>] phys_pmd_init+0x210/0x255
> >>> [ 8321.927724] [<ffffffff815ef2c7>] phys_pud_init+0x1da/0x247
> >>> [ 8321.933951] [<ffffffff815efb81>] kernel_physical_mapping_init+0xf5/0x1d4
> >>> [ 8321.941533] [<ffffffff815ebc7d>] init_memory_mapping+0x18d/0x380
> >>> [ 8321.948341] [<ffffffff810647f9>] arch_add_memory+0x59/0xf0
> >>> [ 8321.954570] [<ffffffff815eceed>] add_memory_resource+0x8d/0x160
> >>> [ 8321.961283] [<ffffffff815ecff2>] add_memory+0x32/0xf0
> >>> [ 8321.967025] [<ffffffff813e1c91>] acpi_memory_device_add+0x131/0x2e0
> >>> [ 8321.974128] [<ffffffff8139f752>] acpi_bus_attach+0xe2/0x190
> >>> [ 8321.980453] [<ffffffff8139f6ce>] acpi_bus_attach+0x5e/0x190
> >>> [ 8321.986778] [<ffffffff8139f6ce>] acpi_bus_attach+0x5e/0x190
> >>> [ 8321.993103] [<ffffffff8139f6ce>] acpi_bus_attach+0x5e/0x190
> >>> [ 8321.999428] [<ffffffff813a1157>] acpi_bus_scan+0x37/0x70
> >>> [ 8322.005461] [<ffffffff81fba955>] acpi_scan_init+0x77/0x1b4
> >>> [ 8322.011690] [<ffffffff81fba70c>] acpi_init+0x297/0x2b3
> >>> [ 8322.017530] [<ffffffff8100213a>] do_one_initcall+0xca/0x1f0
> >>> [ 8322.023855] [<ffffffff81f74266>] kernel_init_freeable+0x194/0x226
> >>> [ 8322.030760] [<ffffffff815eb1ba>] kernel_init+0xa/0xe0
> >>> [ 8322.036503] [<ffffffff815f7bc5>] ret_from_fork+0x55/0x80
> >>>
> >>> So this patch would break it again.
> >>>
> >>> I'd recommend ...
> >>>
> >>>>
> >>>>> Fixes: 357b4da50a62 ("x86: respect memory size limiting via mem= parameter")
> >>>>> Signed-off-by: Baoquan He <bhe@redhat.com>
> >>>>> ---
> >>>>> mm/memory_hotplug.c | 2 +-
> >>>>> 1 file changed, 1 insertion(+), 1 deletion(-)
> >>>>>
> >>>>> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> >>>>> index 55ac23ef11c1..5466a0a00901 100644
> >>>>> --- a/mm/memory_hotplug.c
> >>>>> +++ b/mm/memory_hotplug.c
> >>>>> @@ -105,7 +105,7 @@ static struct resource *register_memory_resource(u64 start, u64 size)
> >>>>> unsigned long flags = IORESOURCE_SYSTEM_RAM | IORESOURCE_BUSY;
> >>>>> char *resource_name = "System RAM";
> >>>>>
> >>>>> - if (start + size > max_mem_size)
> >>>>> + if (start + size > max_mem_size && system_state == SYSTEM_BOOTING)
> >>>
> >>> ... changing this to: ... && system_state != SYSTEM_RUNNING
> >>
> >> I think we usually use system_state < SYSTEM_RUNNING
> >>
> >
> > Works for me as well. :-)
> >

Thanks for reviewing and suggestions, will correct it as
'system_state < SYSTEM_RUNNING'.

>
> As this patch has to be resent, I'd also enjoy a comment explaining why
> this special check is in place
>
> /* Make sure memory hotplug works although mem= was specified */
>
> or sth. like that :)

OK, will consider what is better to be placed here. Thanks.

\
 
 \ /
  Last update: 2019-12-10 09:04    [W:0.067 / U:5.352 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site