| 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
| ||
|
Message-ID: <20180130101141.GW21609@dhcp22.suse.cz>
Date: Tue, 30 Jan 2018 11:11:41 +0100
From: Michal Hocko <mhocko@...nel.org>
To: Bharata B Rao <bharata@...ux.vnet.ibm.com>
Cc: linux-kernel@...r.kernel.org, linux-mm@...ck.org,
pasha.tatashin@...cle.com,
Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: Memory hotplug not increasing the total RAM
[Cc Andrew - thread starts here
http://lkml.kernel.org/r/20180130083006.GB1245@in.ibm.com]
On Tue 30-01-18 15:23:45, Bharata B Rao wrote:
> On Tue, Jan 30, 2018 at 10:28:15AM +0100, Michal Hocko wrote:
> > On Tue 30-01-18 10:16:00, Michal Hocko wrote:
> > > On Tue 30-01-18 14:00:06, Bharata B Rao wrote:
> > > > Hi,
> > > >
> > > > With the latest upstream, I see that memory hotplug is not working
> > > > as expected. The hotplugged memory isn't seen to increase the total
> > > > RAM pages. This has been observed with both x86 and Power guests.
> > > >
> > > > 1. Memory hotplug code intially marks pages as PageReserved via
> > > > __add_section().
> > > > 2. Later the struct page gets cleared in __init_single_page().
> > > > 3. Next online_pages_range() increments totalram_pages only when
> > > > PageReserved is set.
> > >
> > > You are right. I have completely forgot about this late struct page
> > > initialization during onlining. memory hotplug really doesn't want
> > > zeroying. Let me think about a fix.
> >
> > Could you test with the following please? Not an act of beauty but
> > we are initializing memmap in sparse_add_one_section for memory
> > hotplug. I hate how this is different from the initialization case
> > but there is quite a long route to unify those two... So a quick
> > fix should be as follows.
>
> Tested on Power guest, fixes the issue. I can now see the total memory
> size increasing after hotplug.
Thanks for your quick testing. Here we go with the fix.
>From d60b333d4048a84c3172829ec24706c761a7bd44 Mon Sep 17 00:00:00 2001
From: Michal Hocko <mhocko@...e.com>
Date: Tue, 30 Jan 2018 11:02:18 +0100
Subject: [PATCH] mm, memory_hotplug: fix memmap initialization
Bharata has noticed that onlining a newly added memory doesn't increase
the total memory, pointing to f7f99100d8d9 ("mm: stop zeroing memory
during allocation in vmemmap") as a culprit. This commit has changed
the way how the memory for memmaps is initialized and moves it from the
allocation time to the initialization time. This works properly for the
early memmap init path.
It doesn't work for the memory hotplug though because we need to mark
page as reserved when the sparsemem section is created and later
initialize it completely during onlining. memmap_init_zone is called
in the early stage of onlining. With the current code it calls
__init_single_page and as such it clears up the whole stage and
therefore online_pages_range skips those pages.
Fix this by skipping mm_zero_struct_page in __init_single_page for
memory hotplug path. This is quite uggly but unifying both early init
and memory hotplug init paths is a large project. Make sure we plug the
regression at least.
Fixes: f7f99100d8d9 ("mm: stop zeroing memory during allocation in vmemmap")
Cc: stable
Reported-and-Tested-by: Bharata B Rao <bharata@...ux.vnet.ibm.com>
Signed-off-by: Michal Hocko <mhocko@...e.com>
---
mm/page_alloc.c | 22 ++++++++++++++--------
1 file changed, 14 insertions(+), 8 deletions(-)
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 6129f989223a..f548f50c1f3c 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1178,9 +1178,10 @@ static void free_one_page(struct zone *zone,
}
static void __meminit __init_single_page(struct page *page, unsigned long pfn,
- unsigned long zone, int nid)
+ unsigned long zone, int nid, bool zero)
{
- mm_zero_struct_page(page);
+ if (zero)
+ mm_zero_struct_page(page);
set_page_links(page, zone, nid, pfn);
init_page_count(page);
page_mapcount_reset(page);
@@ -1195,9 +1196,9 @@ static void __meminit __init_single_page(struct page *page, unsigned long pfn,
}
static void __meminit __init_single_pfn(unsigned long pfn, unsigned long zone,
- int nid)
+ int nid, bool zero)
{
- return __init_single_page(pfn_to_page(pfn), pfn, zone, nid);
+ return __init_single_page(pfn_to_page(pfn), pfn, zone, nid, zero);
}
#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
@@ -1218,7 +1219,7 @@ static void __meminit init_reserved_page(unsigned long pfn)
if (pfn >= zone->zone_start_pfn && pfn < zone_end_pfn(zone))
break;
}
- __init_single_pfn(pfn, zid, nid);
+ __init_single_pfn(pfn, zid, nid, true);
}
#else
static inline void init_reserved_page(unsigned long pfn)
@@ -1535,7 +1536,7 @@ static unsigned long __init deferred_init_pages(int nid, int zid,
} else {
page++;
}
- __init_single_page(page, pfn, zid, nid);
+ __init_single_page(page, pfn, zid, nid, true);
nr_pages++;
}
return (nr_pages);
@@ -5400,15 +5401,20 @@ void __meminit memmap_init_zone(unsigned long size, int nid, unsigned long zone,
* can be created for invalid pages (for alignment)
* check here not to call set_pageblock_migratetype() against
* pfn out of zone.
+ *
+ * Please note that MEMMAP_HOTPLUG path doesn't clear memmap
+ * because this is done early in sparse_add_one_section
*/
if (!(pfn & (pageblock_nr_pages - 1))) {
struct page *page = pfn_to_page(pfn);
- __init_single_page(page, pfn, zone, nid);
+ __init_single_page(page, pfn, zone, nid,
+ context != MEMMAP_HOTPLUG);
set_pageblock_migratetype(page, MIGRATE_MOVABLE);
cond_resched();
} else {
- __init_single_pfn(pfn, zone, nid);
+ __init_single_pfn(pfn, zone, nid,
+ context != MEMMAP_HOTPLUG);
}
}
}
--
2.15.1
--
Michal Hocko
SUSE Labs
Powered by blists - more mailing lists