[patch] coff-i386, guard against null

[patch] coff-i386, guard against null

[msnyder]

[-- Attachment #1: Type: text/plain, Size: 90 bytes --]

Other code in this function checks to see if sym is null.
If it's null here, it'll fail.


[-- Attachment #2: howto.txt --]
[-- Type: text/plain, Size: 821 bytes --]

2007-07-25  Michael Snyder  <msnyder@access-company.com>

	* coff-i386.c (coff_i386_rtype_to_howto): Guard against null.

Index: coff-i386.c
===================================================================
RCS file: /cvs/src/src/bfd/coff-i386.c,v
retrieving revision 1.26
diff -p -r1.26 coff-i386.c
*** coff-i386.c	3 Jul 2007 14:26:39 -0000	1.26
--- coff-i386.c	26 Jul 2007 00:18:04 -0000
*************** coff_i386_rtype_to_howto (abfd, sec, rel
*** 518,524 ****
        *addendp -= pe_data(sec->output_section->owner)->pe_opthdr.ImageBase;
      }
  
!   if (rel->r_type == R_SECREL32)
      {
        bfd_vma osect_vma;
  
--- 518,524 ----
        *addendp -= pe_data(sec->output_section->owner)->pe_opthdr.ImageBase;
      }
  
!   if (rel->r_type == R_SECREL32 && sym != NULL)
      {
        bfd_vma osect_vma;
  

Re: [patch] coff-i386, guard against null

[H.J. Lu]

On Wed, Jul 25, 2007 at 05:21:13PM -0700, msnyder@sonic.net wrote:
> Other code in this function checks to see if sym is null.
> If it's null here, it'll fail.
> 

> 2007-07-25  Michael Snyder  <msnyder@access-company.com>
> 
> 	* coff-i386.c (coff_i386_rtype_to_howto): Guard against null.
> 

I prefer BFD_ASSERT (sym != NULL).


H.J.

Re: [patch] coff-i386, guard against null

[Daniel Jacobowitz]

On Wed, Jul 25, 2007 at 07:18:55PM -0700, H.J. Lu wrote:
> On Wed, Jul 25, 2007 at 05:21:13PM -0700, msnyder@sonic.net wrote:
> > Other code in this function checks to see if sym is null.
> > If it's null here, it'll fail.
> > 
> 
> > 2007-07-25  Michael Snyder  <msnyder@access-company.com>
> > 
> > 	* coff-i386.c (coff_i386_rtype_to_howto): Guard against null.
> > 
> 
> I prefer BFD_ASSERT (sym != NULL).

BFD_ASSERT returns, though.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [patch] coff-i386, guard against null

[Alan Modra]

On Wed, Jul 25, 2007 at 05:21:13PM -0700, msnyder@sonic.net wrote:
> 	* coff-i386.c (coff_i386_rtype_to_howto): Guard against null.

OK.

-- 
Alan Modra
Australia Development Lab, IBM

Re: [patch] coff-i386, guard against null

[H.J. Lu]

On Thu, Jul 26, 2007 at 07:41:39AM -0400, Daniel Jacobowitz wrote:
> On Wed, Jul 25, 2007 at 07:18:55PM -0700, H.J. Lu wrote:
> > On Wed, Jul 25, 2007 at 05:21:13PM -0700, msnyder@sonic.net wrote:
> > > Other code in this function checks to see if sym is null.
> > > If it's null here, it'll fail.
> > > 
> > 
> > > 2007-07-25  Michael Snyder  <msnyder@access-company.com>
> > > 
> > > 	* coff-i386.c (coff_i386_rtype_to_howto): Guard against null.
> > > 
> > 
> > I prefer BFD_ASSERT (sym != NULL).
> 
> BFD_ASSERT returns, though.

I'd like to know when this condition happens. Crash is better than
silent return.  That is how BFD_ASSERT is used other places.


H.J.

Re: [patch] coff-i386, guard against null

[msnyder]

> On Thu, Jul 26, 2007 at 07:41:39AM -0400, Daniel Jacobowitz wrote:
>> On Wed, Jul 25, 2007 at 07:18:55PM -0700, H.J. Lu wrote:
>> > On Wed, Jul 25, 2007 at 05:21:13PM -0700, msnyder@sonic.net wrote:
>> > > Other code in this function checks to see if sym is null.
>> > > If it's null here, it'll fail.
>> > >
>> >
>> > > 2007-07-25  Michael Snyder  <msnyder@access-company.com>
>> > >
>> > > 	* coff-i386.c (coff_i386_rtype_to_howto): Guard against null.
>> > >
>> >
>> > I prefer BFD_ASSERT (sym != NULL).
>>
>> BFD_ASSERT returns, though.
>
> I'd like to know when this condition happens. Crash is better than
> silent return.  That is how BFD_ASSERT is used other places.

But it doesn't fix the problem that I set out to fix.
If it returns, we'll still crash.

H.J., the change I submitted is consistant with existing code
in this module.  There are six local uses of "if (x != NULL)",
and only one local use of BFD_ASSERT.  I don't mind if you want
to add a BFD_ASSERT in addition, but why not let my change go in?

Re: [patch] coff-i386, guard against null

[H.J. Lu]

On Thu, Jul 26, 2007 at 11:52:07AM -0700, msnyder@sonic.net wrote:
> > On Thu, Jul 26, 2007 at 07:41:39AM -0400, Daniel Jacobowitz wrote:
> >> On Wed, Jul 25, 2007 at 07:18:55PM -0700, H.J. Lu wrote:
> >> > On Wed, Jul 25, 2007 at 05:21:13PM -0700, msnyder@sonic.net wrote:
> >> > > Other code in this function checks to see if sym is null.
> >> > > If it's null here, it'll fail.
> >> > >
> >> >
> >> > > 2007-07-25  Michael Snyder  <msnyder@access-company.com>
> >> > >
> >> > > 	* coff-i386.c (coff_i386_rtype_to_howto): Guard against null.
> >> > >
> >> >
> >> > I prefer BFD_ASSERT (sym != NULL).
> >>
> >> BFD_ASSERT returns, though.
> >
> > I'd like to know when this condition happens. Crash is better than
> > silent return.  That is how BFD_ASSERT is used other places.
> 
> But it doesn't fix the problem that I set out to fix.
> If it returns, we'll still crash.
> 
> H.J., the change I submitted is consistant with existing code
> in this module.  There are six local uses of "if (x != NULL)",
> and only one local use of BFD_ASSERT.  I don't mind if you want
> to add a BFD_ASSERT in addition, but why not let my change go in?

OK with BFD_ASSERT.


H.J.

Re: [patch] coff-i386, guard against null

[msnyder]

> On Thu, Jul 26, 2007 at 11:52:07AM -0700, msnyder@sonic.net wrote:
>> > On Thu, Jul 26, 2007 at 07:41:39AM -0400, Daniel Jacobowitz wrote:
>> >> On Wed, Jul 25, 2007 at 07:18:55PM -0700, H.J. Lu wrote:
>> >> > On Wed, Jul 25, 2007 at 05:21:13PM -0700, msnyder@sonic.net wrote:
>> >> > > Other code in this function checks to see if sym is null.
>> >> > > If it's null here, it'll fail.
>> >> > >
>> >> >
>> >> > > 2007-07-25  Michael Snyder  <msnyder@access-company.com>
>> >> > >
>> >> > > 	* coff-i386.c (coff_i386_rtype_to_howto): Guard against null.
>> >> > >
>> >> >
>> >> > I prefer BFD_ASSERT (sym != NULL).
>> >>
>> >> BFD_ASSERT returns, though.
>> >
>> > I'd like to know when this condition happens. Crash is better than
>> > silent return.  That is how BFD_ASSERT is used other places.
>>
>> But it doesn't fix the problem that I set out to fix.
>> If it returns, we'll still crash.
>>
>> H.J., the change I submitted is consistant with existing code
>> in this module.  There are six local uses of "if (x != NULL)",
>> and only one local use of BFD_ASSERT.  I don't mind if you want
>> to add a BFD_ASSERT in addition, but why not let my change go in?
>
> OK with BFD_ASSERT.

Sorry, I don't understand your reply.

Re: [patch] coff-i386, guard against null

[H.J. Lu]

On Thu, Jul 26, 2007 at 04:24:22PM -0700, msnyder@sonic.net wrote:
> > On Thu, Jul 26, 2007 at 11:52:07AM -0700, msnyder@sonic.net wrote:
> >> > On Thu, Jul 26, 2007 at 07:41:39AM -0400, Daniel Jacobowitz wrote:
> >> >> On Wed, Jul 25, 2007 at 07:18:55PM -0700, H.J. Lu wrote:
> >> >> > On Wed, Jul 25, 2007 at 05:21:13PM -0700, msnyder@sonic.net wrote:
> >> >> > > Other code in this function checks to see if sym is null.
> >> >> > > If it's null here, it'll fail.
> >> >> > >
> >> >> >
> >> >> > > 2007-07-25  Michael Snyder  <msnyder@access-company.com>
> >> >> > >
> >> >> > > 	* coff-i386.c (coff_i386_rtype_to_howto): Guard against null.
> >> >> > >
> >> >> >
> >> >> > I prefer BFD_ASSERT (sym != NULL).
> >> >>
> >> >> BFD_ASSERT returns, though.
> >> >
> >> > I'd like to know when this condition happens. Crash is better than
> >> > silent return.  That is how BFD_ASSERT is used other places.
> >>
> >> But it doesn't fix the problem that I set out to fix.
> >> If it returns, we'll still crash.
> >>
> >> H.J., the change I submitted is consistant with existing code
> >> in this module.  There are six local uses of "if (x != NULL)",
> >> and only one local use of BFD_ASSERT.  I don't mind if you want
> >> to add a BFD_ASSERT in addition, but why not let my change go in?
> >
> > OK with BFD_ASSERT.
> 
> Sorry, I don't understand your reply.

Your change is OK if you also add a BFD_ASSERT so that we will
know something is wrong.

Thanks.

H.J.

Re: [patch] coff-i386, guard against null

[msnyder]

> On Thu, Jul 26, 2007 at 04:24:22PM -0700, msnyder@sonic.net wrote:
>> > On Thu, Jul 26, 2007 at 11:52:07AM -0700, msnyder@sonic.net wrote:
>> >> > On Thu, Jul 26, 2007 at 07:41:39AM -0400, Daniel Jacobowitz wrote:
>> >> >> On Wed, Jul 25, 2007 at 07:18:55PM -0700, H.J. Lu wrote:
>> >> >> > On Wed, Jul 25, 2007 at 05:21:13PM -0700, msnyder@sonic.net
>> wrote:
>> >> >> > > Other code in this function checks to see if sym is null.
>> >> >> > > If it's null here, it'll fail.
>> >> >> > >
>> >> >> >
>> >> >> > > 2007-07-25  Michael Snyder  <msnyder@access-company.com>
>> >> >> > >
>> >> >> > > 	* coff-i386.c (coff_i386_rtype_to_howto): Guard against null.
>> >> >> > >
>> >> >> >
>> >> >> > I prefer BFD_ASSERT (sym != NULL).
>> >> >>
>> >> >> BFD_ASSERT returns, though.
>> >> >
>> >> > I'd like to know when this condition happens. Crash is better than
>> >> > silent return.  That is how BFD_ASSERT is used other places.
>> >>
>> >> But it doesn't fix the problem that I set out to fix.
>> >> If it returns, we'll still crash.
>> >>
>> >> H.J., the change I submitted is consistant with existing code
>> >> in this module.  There are six local uses of "if (x != NULL)",
>> >> and only one local use of BFD_ASSERT.  I don't mind if you want
>> >> to add a BFD_ASSERT in addition, but why not let my change go in?
>> >
>> > OK with BFD_ASSERT.
>>
>> Sorry, I don't understand your reply.
>
> Your change is OK if you also add a BFD_ASSERT so that we will
> know something is wrong.

OK, committed.
Thanks.