linux-kernel - Re: [RFC][PATCH 00/13] Cleanup recordmcount and begin objtool conversion

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] [day] [month] [year] [list]

Message-ID: <20190531183451.vncrseq2s7cpevrp@treble>
Date:   Fri, 31 May 2019 13:34:51 -0500
From:   Josh Poimboeuf <jpoimboe@...hat.com>
To:     Matt Helsley <mhelsley@...are.com>
Cc:     Peter Zijlstra <peterz@...radead.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Ingo Molnar <mingo@...nel.org>,
        Steven Rostedt <rostedt@...dmis.org>
Subject: Re: [RFC][PATCH 00/13] Cleanup recordmcount and begin objtool
 conversion

On Thu, May 30, 2019 at 04:52:19PM -0700, Matt Helsley wrote:
> > > objtool orc+mcount ?
> > > 
> > > That is, have '+' be a separator for cmd thingies. That would of course
> > > require all other arguments to be shared between all commands, which is
> > > currently already so, but I've not checked the mcount patches.
> > 
> > The problem is that you have to combine "orc generate" with "mcount
> > record".  Because even the subcommands have subcommands ;-)
> > 
> > And also sharing arguments between all subcommands isn't ideal.
> > 
> > Maybe could do:
> > 
> >   objtool orc generate [orc options] + mcount record [mcount options]
> 
> I think that makes more sense; it'll be easier to construct
> Make recipes this way. I was thinking '+' would be something like the
> getopt handling of the '--' argument where it stops argument parsing so
> someting else can consume the remainder.
> 
> The really interesting part is deciding which file to operate on is
> specified by the arguments to the first subcommand and subsequent subcmds
> would then operate on the same object file. For example:
> 
> objtool orc generate [orc opts] foo.o + mcount record [mcount opts]
> 
> Would it be clearer what's going on if the object file(s) were specified
> first and then the passes to run and their arguments came afterwards?
> I'm thinking it'd go somewhat like this:
> 
> objtool foo.o [bar.o] -- check [check opts] + \
> 			 orc generate [orc opts] + \
> 			 mcount record [mcount opts]
> 
> Then objtool would iterate over the object file(s) to open,
> hand off the ELF data structures into each successive pass, and
> finally write any accumulated changes back.

Yeah, I forgot about the .o file.  Something like that would probably
work.  The ordering seems a bit funny to me.  Another possibly more
readable variation would be:

  objtool check [check opts] + orc generate [orc opts] + mcount record [mcount opts] -- foo.o [bar.o]

or, just use '--' as a generic separator which can be used to separate
subcommands or file names.

  objtool check [check opts] -- orc generate [orc opts] -- mcount record [mcount opts] -- foo.o [bar.o]

I kind of like that.  But I think any of these variations would probably
work.

-- 
Josh