linux-kernel - Re: [PATCH 02/18] objtool: Support data symbol printing

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, 14 Apr 2022 08:38:54 -0700
From:   Josh Poimboeuf <jpoimboe@...hat.com>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     x86@...nel.org, linux-kernel@...r.kernel.org,
        Miroslav Benes <mbenes@...e.cz>
Subject: Re: [PATCH 02/18] objtool: Support data symbol printing

On Thu, Apr 14, 2022 at 05:31:57PM +0200, Peter Zijlstra wrote:
> On Thu, Apr 14, 2022 at 08:21:48AM -0700, Josh Poimboeuf wrote:
> > On Thu, Apr 14, 2022 at 09:05:47AM +0200, Peter Zijlstra wrote:
> > > On Wed, Apr 13, 2022 at 04:19:37PM -0700, Josh Poimboeuf wrote:
> > > 
> > > > @@ -34,8 +37,8 @@ static inline char *offstr(struct section *sec, unsigned long offset)
> > > >  
> > > >  	str = malloc(strlen(name) + 20);
> > > >  
> > > > -	if (func)
> > > > -		sprintf(str, "%s()+0x%lx", name, name_off);
> > > > +	if (sym)
> > > > +		sprintf(str, "%s%s+0x%lx", name, is_text ? "()" : "", name_off);
> > > >  	else
> > > >  		sprintf(str, "%s+0x%lx", name, name_off);
> > > 
> > > So I like the patch, except that "()" thing is something where we differ
> > > from the kernel's %ps format and I've cursed it a number of times
> > > because I then have to manually edit (iow remove) things when pasting it
> > > in various scripts etc..
> > 
> > Oh, hm, that's true.  I can remove them if you prefer.
> 
> Yeah, I think taking it out is best, easier if we're consistent with %ps
> for everybody.

Will do.

> > > That said, it totally makes sense to differentiate between a text and
> > > data symbol this way :/
> > 
> > Yes, but if we're keeping the "Add sec+offset to warnings" patch then
> > that distinction is already (kind of) being made by showing the data
> > section name.  And the data symbol warnings should be rare.
> 
> Yes, I'd not seen that yet, what's that for? The Changelog alludes to
> something, but I don't think it actually does get used later.

Nick had asked for something like that, it's just a way to avoid doing
math every time we look at a warning, i.e. to convert func+offset to
sec+offset.

But it's kind of ugly and I'm not 100% happy with it.

Maybe it should be behind an option (--sec-offsets)?

-- 
Josh