linux-kernel - Re: [PATCH v2 5/5] erofs: use meta buffers for zmap operations

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:   Tue, 4 Jan 2022 17:02:00 +0800
From:   Gao Xiang <hsiangkao@...ux.alibaba.com>
To:     Chao Yu <chao@...nel.org>
Cc:     linux-erofs@...ts.ozlabs.org, Liu Bo <bo.liu@...ux.alibaba.com>,
        LKML <linux-kernel@...r.kernel.org>, Yue Hu <huyue2@...ong.com>
Subject: Re: [PATCH v2 5/5] erofs: use meta buffers for zmap operations

Hi Chao,

On Tue, Jan 04, 2022 at 04:15:05PM +0800, Chao Yu wrote:
> On 2022/1/2 12:00, Gao Xiang wrote:

...

> > --- a/fs/erofs/zdata.c
> > +++ b/fs/erofs/zdata.c
> > @@ -698,20 +698,18 @@ static int z_erofs_do_read_page(struct z_erofs_decompress_frontend *fe,
> >   		goto err_out;
> >   	if (z_erofs_is_inline_pcluster(clt->pcl)) {
> > -		struct page *mpage;
> > +		void *mp;
> > -		mpage = erofs_get_meta_page(inode->i_sb,
> > -					    erofs_blknr(map->m_pa));
> > -		if (IS_ERR(mpage)) {
> > -			err = PTR_ERR(mpage);
> > +		mp = erofs_read_metabuf(&fe->map.buf, inode->i_sb,
> > +					erofs_blknr(map->m_pa), EROFS_NO_KMAP);
> > +		if (IS_ERR(mp)) {
> > +			err = PTR_ERR(mp);
> >   			erofs_err(inode->i_sb,
> >   				  "failed to get inline page, err %d", err);
> >   			goto err_out;
> >   		}
> > -		/* TODO: new subpage feature will get rid of it */
> > -		unlock_page(mpage);
> > -
> > -		WRITE_ONCE(clt->pcl->compressed_pages[0], mpage);
> > +		get_page(fe->map.buf.page);
> 
> Comparing to previous implementation, it adds an extra reference on the page, why?

Thanks for the question. Previously, erofs_get_meta_page was called
independently without reusing zmap mpage, so the page refcount had no
relationship with zmap mpage.

However, now we reuse zmap metabuf instead(fe->map.buf), so an extra
page refcount is needed since zmap metabuf will be released at the end
of readpage or readahead...

Thanks,
Gao Xiang