linux-kernel - Re: [PATCH 8/8] Per-container pages reclamation

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 17 May 2007 17:01:26 +0530
From:	Balbir Singh <balbir@...ux.vnet.ibm.com>
To:	Pavel Emelianov <xemul@...ru>
CC:	Andrew Morton <akpm@...l.org>, Paul Menage <menage@...gle.com>,
	Srivatsa Vaddagiri <vatsa@...ibm.com>,
	Balbir Singh <balbir@...ibm.com>, devel@...nvz.org,
	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
	Kirill Korotaev <dev@...ru>,
	Chandra Seetharaman <sekharan@...ibm.com>,
	Cedric Le Goater <clg@...ibm.com>,
	"Eric W. Biederman" <ebiederm@...ssion.com>,
	Rohit Seth <rohitseth@...gle.com>,
	Linux Containers <containers@...ts.osdl.org>,
	Linux Memory Management List <linux-mm@...ck.org>
Subject: Re: [PATCH 8/8] Per-container pages reclamation

Pavel Emelianov wrote:
> Implement try_to_free_pages_in_container() to free the
> pages in container that has run out of memory.
> 
> The scan_control->isolate_pages() function isolates the
> container pages only.
> 
> 

Hi, Pavel/Andrew,

I've started running some basic tests like lmbench and LTP vm stress
on the RSS controller.

With the controller rss_limit set to 256 MB, I saw the following panic
on a machine

Unable to handle kernel NULL pointer dereference at 000000000000001c RIP: 
 [<ffffffff80328581>] _raw_spin_lock+0xd/0xf6
PGD 3c841067 PUD 5d5d067 PMD 0 
Oops: 0000 [1] SMP 
CPU 2 
Modules linked in: ipv6 hidp rfcomm l2cap bluetooth sunrpc video button battery asus_acpi backlight ac lp parport_pc parport nvram pcspkr amd_rng rng_core i2c_amd756 i2c_core
Pid: 13581, comm: mtest01 Not tainted 2.6.20-autokern1 #1
RIP: 0010:[<ffffffff80328581>]  [<ffffffff80328581>] _raw_spin_lock+0xd/0xf6
RSP: 0000:ffff81003e6c9ce8  EFLAGS: 00010096
RAX: ffffffff8087f720 RBX: 0000000000000018 RCX: ffff81003f36f9d0
RDX: ffff8100807bb040 RSI: 0000000000000001 RDI: 0000000000000018
RBP: 0000000000000000 R08: ffff81003e6c8000 R09: 0000000000000002
R10: ffff810001021da8 R11: ffffffff8044658f R12: ffff81000c861e01
R13: 0000000000000018 R14: ffff81000c861eb8 R15: ffff810032d34138
FS:  00002abf7a1961e0(0000) GS:ffff81003edb94c0(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 000000000000001c CR3: 000000002ba6e000 CR4: 00000000000006e0
Process mtest01 (pid: 13581, threadinfo ffff81003e6c8000, task ffff81003d8ec040)
Stack:  ffff810001003638 ffff810014a8c2c0 0000000000000000 ffff81000c861e01
 0000000000000018 ffffffff80287166 ffff81000c861eb8 ffff81000000bac0
 ffff81003f36f9a0 ffff81000c861e40 ffff81001d4b6a20 ffffffff8026a92e
Call Trace:
 [<ffffffff80287166>] container_rss_move_lists+0x3b/0xaf
 [<ffffffff8026a92e>] activate_page+0xc1/0xd0
 [<ffffffff80245f15>] wake_bit_function+0x0/0x23
 [<ffffffff8026ab34>] mark_page_accessed+0x1b/0x2f
 [<ffffffff80265d25>] filemap_nopage+0x180/0x338
 [<ffffffff80270474>] __handle_mm_fault+0x1f2/0xa81
 [<ffffffff804c58ef>] do_page_fault+0x42b/0x7b3
 [<ffffffff802484c4>] hrtimer_cancel+0xc/0x16
 [<ffffffff804c2a89>] do_nanosleep+0x47/0x70
 [<ffffffff802485f4>] hrtimer_nanosleep+0x58/0x119
 [<ffffffff8023bc1f>] sys_sysinfo+0x15b/0x173
 [<ffffffff804c3d3d>] error_exit+0x0/0x84

On analyzing the code, I found that the page is mapped (we have a page_mapped() check in
container_rss_move_lists()), but the page_container is invalid. Please review the fix
attached (we reset the page's container pointer to NULL when a page is completely unmapped)

-- 
	Warm Regards,
	Balbir Singh
	Linux Technology Center
	IBM, ISTL

View attachment "rss-fix-lru-race.patch" of type "text/x-patch" (2195 bytes)