xen-devel.lists.xenproject.org archive mirror
 help / color / mirror / Atom feed
From: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
To: Jan Beulich <JBeulich@suse.com>
Cc: Stefano Stabellini <sstabellini@kernel.org>,
	ross.lagerwall@citrix.com,
	Andrew Cooper <andrew.cooper3@citrix.com>,
	Julien Grall <julien.grall@arm.com>,
	xen-devel@lists.xenproject.org
Subject: Re: [PATCH v4 6/9] livepatch: Add parsing for the symbol+0x<offset>
Date: Tue, 6 Sep 2016 15:56:08 -0400	[thread overview]
Message-ID: <20160906195608.GA32099@char.us.oracle.com> (raw)
In-Reply-To: <57BD8011020000780010883A@prv-mh.provo.novell.com>

On Wed, Aug 24, 2016 at 03:08:01AM -0600, Jan Beulich wrote:
> >>> On 24.08.16 at 04:22, <konrad.wilk@oracle.com> wrote:
> > --- a/xen/common/livepatch.c
> > +++ b/xen/common/livepatch.c
> > @@ -237,13 +237,34 @@ static const char *livepatch_symbols_lookup(unsigned long addr,
> >  static int resolve_old_address(struct livepatch_func *f,
> >                                 const struct livepatch_elf *elf)
> >  {
> > +    const char *s;
> > +    char *plus = NULL;
> 
> Pointless initializer.

We need that otherwise this part (which is at the bottom of this function):

    if ( plus )
    {
        *plus = '+';
        f->old_addr += offset;
    }


May be invoked for symbols that that don't have the '+' in them.
> 
> > +    unsigned long offset = 0;
> > +
> >      if ( f->old_addr )
> >          return 0;
> >  
> > -    f->old_addr = (void *)symbols_lookup_by_name(f->name);
> > +    s = f->name;
> 
> Otoh this could become s'es initializer.

<nods>
> 
> > +    /* +<offset> */
> > +    plus = strchr(f->name, '+');
> 
> And I think you should prefer using the local variable here.

<nods>

> 
> Furthermore you're losing const here - does f->name really point
> to memory that doesn't get mapped r/o?

Yes.

The 'struct livepatch_func' contains the the ->opaque array of 31 bytes
(from which we use 5 bytes) which the hypervisor uses to stash the original
instructions.
> 
> > +    if ( plus )
> > +    {
> > +        const char *endp = NULL;
> 
> Pointless initializer again (or else ...
> 
> > +        offset = simple_strtoul(plus + 1, &endp, 16);
> > +
> > +        if ( *endp != '\0' )
> 
> ... the deref here couldn't be unconditional).
> 
> > +            return -EINVAL;
> > +
> > +        /* So that symbol lookup works. */
> > +        *plus = '\0';
> > +        s = f->name;
> 
> Why? f->name didn't change afaict.

Ugh. Stale code.
> 
> Overall - are you sure you want to disallow symbol names containing
> + characters? I.e. you don't want to add support for some form of
> quoting?

Can you actually have + in a function or object?

Or are you asking in terms of the '\0' overwritting the '+' ?
We put it back later on so that the symbol still has + in it:

 274     if ( plus )                                                                 
 275     {                                                                           
 276         *plus = '+';                                                            
 277         f->old_addr += offset;                                                  
 278     }                                           

I will change the comment from 'So that symbol lookup works.' to
'Temporary mutilation of name so that symbol lookup works.'

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel

  reply	other threads:[~2016-09-06 19:56 UTC|newest]

Thread overview: 40+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-08-24  2:22 [PATCH v4] Livepatch fixes and features for v4.8 Konrad Rzeszutek Wilk
2016-08-24  2:22 ` [PATCH v4 1/9] livepatch: Clear .bss when payload is reverted Konrad Rzeszutek Wilk
2016-08-24  8:55   ` Jan Beulich
2016-08-25 16:08     ` Andrew Cooper
2016-09-06 16:51       ` Konrad Rzeszutek Wilk
2016-09-07  9:18         ` Jan Beulich
2016-09-06 16:47     ` Konrad Rzeszutek Wilk
2016-09-07  8:02       ` Jan Beulich
2016-09-08  9:25         ` Konrad Rzeszutek Wilk
2016-09-09 13:33   ` Ross Lagerwall
2016-09-09 13:50     ` Konrad Rzeszutek Wilk
2016-09-09 13:58       ` Ross Lagerwall
2016-09-09 15:28       ` Jan Beulich
2016-08-24  2:22 ` [PATCH v4 2/9] livepatch: Deal with payloads without any .text Konrad Rzeszutek Wilk
2016-08-24  2:22 ` [PATCH v4 3/9] version/livepatch: Move xen_build_id_check to version.h Konrad Rzeszutek Wilk
2016-08-24  2:22 ` [PATCH v4 4/9] version: Print build-id at bootup Konrad Rzeszutek Wilk
2016-08-24  8:58   ` Jan Beulich
2016-09-06 16:57     ` Konrad Rzeszutek Wilk
2016-09-07  8:03       ` Jan Beulich
2016-09-09 13:37       ` Ross Lagerwall
2016-08-24  2:22 ` [PATCH v4 5/9] livepatch: Move code from prepare_payload to own routine Konrad Rzeszutek Wilk
2016-08-25 16:02   ` Ross Lagerwall
2016-08-24  2:22 ` [PATCH v4 6/9] livepatch: Add parsing for the symbol+0x<offset> Konrad Rzeszutek Wilk
2016-08-24  9:08   ` Jan Beulich
2016-09-06 19:56     ` Konrad Rzeszutek Wilk [this message]
2016-09-07  8:10       ` Jan Beulich
2016-09-08  9:22         ` Konrad Rzeszutek Wilk
2016-09-08 10:01           ` Jan Beulich
2016-09-09 14:28           ` Ross Lagerwall
2016-08-24  2:22 ` [PATCH v4 7/9] livepatch: NOP if func->new_[addr] is zero Konrad Rzeszutek Wilk
2016-08-24  9:13   ` Jan Beulich
2016-09-06 20:05     ` Konrad Rzeszutek Wilk
2016-09-07  8:13       ` Jan Beulich
2016-08-24  2:22 ` [PATCH v4 8/9] symbols: Generate an xen-sym.map Konrad Rzeszutek Wilk
2016-08-24  9:16   ` Jan Beulich
2016-09-09 13:43   ` Ross Lagerwall
2016-08-24  2:22 ` [PATCH v4 9/9] livepach: Add .livepatch.hooks functions and test-case Konrad Rzeszutek Wilk
2016-09-06 17:22   ` Konrad Rzeszutek Wilk
2016-09-06 18:25     ` Andrew Cooper
2016-09-08  1:18       ` Konrad Rzeszutek Wilk

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20160906195608.GA32099@char.us.oracle.com \
    --to=konrad.wilk@oracle.com \
    --cc=JBeulich@suse.com \
    --cc=andrew.cooper3@citrix.com \
    --cc=julien.grall@arm.com \
    --cc=ross.lagerwall@citrix.com \
    --cc=sstabellini@kernel.org \
    --cc=xen-devel@lists.xenproject.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).