lkml.org 
[lkml]   [2017]   [Nov]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH v2 1/1] mm: buddy page accessed before initialized
On Thu 02-11-17 13:02:21, Pavel Tatashin wrote:
> This problem is seen when machine is rebooted after kexec:
> A message like this is printed:
> ==========================================================================
> WARNING: CPU: 21 PID: 249 at linux/lib/list_debug.c:53__listd+0x83/0xa0
> Modules linked in:
> CPU: 21 PID: 249 Comm: pgdatinit0 Not tainted 4.14.0-rc6_pt_deferred #90
> Hardware name: Oracle Corporation ORACLE SERVER X6-2/ASM,MOTHERBOARD,1U,
> BIOS 3016
> node 1 initialised, 32444607 pages in 1679ms
> task: ffff880180e75a00 task.stack: ffffc9000cdb0000
> RIP: 0010:__list_del_entry_valid+0x83/0xa0
> RSP: 0000:ffffc9000cdb3d18 EFLAGS: 00010046
> RAX: 0000000000000054 RBX: 0000000000000009 RCX: ffffffff81c5f3e8
> RDX: 0000000000000000 RSI: 0000000000000086 RDI: 0000000000000046
> RBP: ffffc9000cdb3d18 R08: 00000000fffffffe R09: 0000000000000154
> R10: 0000000000000005 R11: 0000000000000153 R12: 0000000001fcdc00
> R13: 0000000001fcde00 R14: ffff88207ffded00 R15: ffffea007f370000
> FS: 0000000000000000(0000) GS:ffff881fffac0000(0000) knlGS:0
> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 0000000000000000 CR3: 000000407ec09001 CR4: 00000000003606e0
> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> Call Trace:
> free_one_page+0x103/0x390
> __free_pages_ok+0x1cf/0x2d0
> __free_pages+0x19/0x30
> __free_pages_boot_core+0xae/0xba
> deferred_free_range+0x60/0x94
> deferred_init_memmap+0x324/0x372
> kthread+0x109/0x140
> ? __free_pages_bootmem+0x2e/0x2e
> ? kthread_park+0x60/0x60
> ret_from_fork+0x25/0x30
>
> list_del corruption. next->prev should be ffffea007f428020, but was
> ffffea007f1d8020
> ==========================================================================
>
> The problem happens in this path:
>
> page_alloc_init_late
> deferred_init_memmap
> deferred_init_range
> __def_free
> deferred_free_range
> __free_pages_boot_core(page, order)
> __free_pages()
> __free_pages_ok()
> free_one_page()
> __free_one_page(page, pfn, zone, order, migratetype);
>
> deferred_init_range() initializes one page at a time by calling
> __init_single_page(), once it initializes pageblock_nr_pages pages, it
> calls deferred_free_range() to free the initialized pages to the buddy
> allocator. Eventually, we reach __free_one_page(), where we compute buddy
> page:
> buddy_pfn = __find_buddy_pfn(pfn, order);
> buddy = page + (buddy_pfn - pfn);
>
> buddy_pfn is computed as pfn ^ (1 << order), or pfn + pageblock_nr_pages.
> Thefore, buddy page becomes a page one after the range that currently was
> initialized, and we access this page in this function. Also, later when we
> return back to deferred_init_range(), the buddy page is initialized again.
>
> So, in order to avoid this issue, we must initialize the buddy page prior
> to calling deferred_free_range().

Have you measured any negative performance impact with this change?

> Signed-off-by: Pavel Tatashin <pasha.tatashin@oracle.com>

The patch looks good to me otherwise. So if this doesn't introduce a
noticeable overhead, which I whope it doesn't then feel free to add
Acked-by: Michal Hocko <mhocko@suse.com>
--
Michal Hocko
SUSE Labs

\
 
 \ /
  Last update: 2017-11-03 10:27    [W:0.067 / U:0.372 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site