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
| ||
|
Date: Wed, 29 Mar 2017 15:51:23 +0800 From: Yisheng Xie <xieyisheng1@...wei.com> To: Minchan Kim <minchan@...nel.org> CC: <ngupta@...are.org>, <sergey.senozhatsky.work@...il.com>, <linux-mm@...ck.org>, Linux Kernel Mailing List <linux-kernel@...r.kernel.org>, Xishi Qiu <qiuxishi@...wei.com>, Hanjun Guo <guohanjun@...wei.com> Subject: Re: [RFC]mm/zsmalloc,: trigger BUG_ON in function zs_map_object. Hi Minchan, Thanks for your comment! On 2017/3/29 8:20, Minchan Kim wrote: > Hello, > > On Tue, Mar 28, 2017 at 03:20:22PM +0800, Yisheng Xie wrote: >> Hi, all, >> >> We had backport the no-lru migration to linux-4.1, meanwhile change the >> ZS_MAX_ZSPAGE_ORDER to 3. Then we met a BUG_ON(!page[1]). > > Hmm, I don't know how you backported. Yes, maybe caused by our unsuitable backport. > > There isn't any problem with default ZS_MAX_ZSPAGE_ORDER. Right? > So, it happens only if you changed it to 3? I will check whether it will default ZS_MAX_ZSPAGE_ORDER. > > Could you tell me what is your base kernel? and what zram/zsmalloc > version(ie, from what kernel version) you backported to your > base kernel? > We backport from kernel v4.8-rc8 to kernel v4.1. >> >> It rarely happen, and presently, what I get is: >> [6823.316528s]obj=a160701f, obj_idx=15, class{size:2176,objs_per_zspage:15,pages_per_zspage:8} >> [...] >> [6823.316619s]BUG: failure at /home/ethan/kernel/linux-4.1/mm/zsmalloc.c:1458/zs_map_object()! ----> BUG_ON(!page[1]) >> >> It seems that we have allocated an object from a ZS_FULL group? >> (Actually, I do not get the inuse number of this zspage, which I am trying to.) >> And presently, I can not find why it happened. Any idea about it? > > Although it happens rarely, always above same symptom once it happens? Yes , though the class size is not the same, which means not from the same class. however, the (obj_idx == objs_per_zspage) is always true. Thanks Yisheng Xie.
Powered by blists - more mailing lists