linux-kernel - Re: [PATCH 06/06] 9p fid refcount: cleanup p9_fid

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: <YqeVFTkREfWUfPFi@codewreck.org>
Date:   Tue, 14 Jun 2022 04:50:45 +0900
From:   Dominique Martinet <asmadeus@...ewreck.org>
To:     Tyler Hicks <tyhicks@...ux.microsoft.com>
Cc:     Christian Schoenebeck <linux_oss@...debyte.com>,
        Eric Van Hensbergen <ericvh@...il.com>,
        Latchesar Ionkov <lucho@...kov.net>,
        v9fs-developer@...ts.sourceforge.net, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 06/06] 9p fid refcount: cleanup p9_fid_put calls

Thanks for the reviews,

Tyler Hicks wrote on Mon, Jun 13, 2022 at 12:55:09PM -0500:
> > @@ -189,13 +197,13 @@ static struct p9_fid *v9fs_fid_lookup_with_uid(struct dentry *dentry,
> >  		else
> >  			uname = v9ses->uname;
> >  
> > -		root_fid = p9_client_attach(v9ses->clnt, NULL, uname, uid,
> > -					    v9ses->aname);
> > -		if (IS_ERR(root_fid))
> > -			return root_fid;
> > +		fid = p9_client_attach(v9ses->clnt, NULL, uname, uid,
> 
> To keep the readability benefits in my "9p: Track the root fid with its
> own variable during lookups" patch, I think root_fid should be assigned
> here and then used in the error check and return statement.
> 
> > +				       v9ses->aname);
> > +		if (IS_ERR(fid))
> > +			return fid;
> >  
> > -		p9_fid_get(root_fid);
> > -		v9fs_fid_add(dentry->d_sb->s_root, root_fid);
> > +		root_fid = p9_fid_get(fid);
> > +		v9fs_fid_add(dentry->d_sb->s_root, &fid);
> 
> root_fid should be used in the two lines above, too.

This actually was the only place where we still want to use the root_fid
after calling v9fs_fid_add; if we keep root_fid above we need to do
something like

fid = p9_fid_get(root_fid);
v9fs_Fid_add(dentry->d_sb->s_root, &root_fid);
root_fid = fid;
// fid = NULL; ? not strictly needed as we set it again shortly afterwards

which I wanted to avoid, but I guess I don't mind strongly either way --
pick your poison.
I could also just keep v9fs_fid_add as a non-stealing version, but I
think it's better that way as it strongly signal that we stashed that
ref away and shouldn't use the fid anymore unless another ref was
obtained through fid_get like we do here.
(I was actually tempted to do the same with p9_fid_put, but checking
other kernel "put"s I didn't see any code that does this so I refrained
from that churn)

-- 
Dominique