linux-kernel - Re: [PATCH 02/10] staging: erofs: fix race when the managed cache is enabled

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]

Message-ID: <20181122132041.GA14300@kroah.com>
Date:   Thu, 22 Nov 2018 14:20:41 +0100
From:   Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To:     Gao Xiang <gaoxiang25@...wei.com>
Cc:     devel@...verdev.osuosl.org, linux-erofs@...ts.ozlabs.org,
        Chao Yu <yuchao0@...wei.com>,
        LKML <linux-kernel@...r.kernel.org>, weidu.du@...wei.com,
        Miao Xie <miaoxie@...wei.com>
Subject: Re: [PATCH 02/10] staging: erofs: fix race when the managed cache is
 enabled

On Thu, Nov 22, 2018 at 08:41:15PM +0800, Gao Xiang wrote:
> Hi Greg,
> 
> On 2018/11/22 20:26, Greg Kroah-Hartman wrote:
> > Ugh, every page?  Ok, nevermind, I take back my objections.  You all are
> > crazy and need to do crazy things like this :)
> 
> ...Do you have some idea about this? ... I think it is fairly normal... :(
> 
> We have a large number of managed workgroups, the current use case is the 4k compressed size,
> each compressed page has a workgroup, but erofs also has reclaim paths to free them for low memory cases.
> 
> But since the structure (z_erofs_vle_workgroup, erofs_workgroup) is critical,
> I need to make it as small as possible.

I do not know "real" filesystems (i.e. ones that put stuff on disks),
only virtual ones (like sysfs/kernfs), so I do not know what to suggest
here, sorry.  I thought this was a much higner-level structure, if this
is per-compressed-page, then you are more than welcome to optimize for
this.  Good luck with your "fake" spinlocks though :)

thanks,

greg k-h