Re: PATCH: Allow sections in adjacent pages to be included in the same segment

["H.J. Lu" <hjl dot tools at gmail dot com>]

On Tue, Jul 22, 2008 at 4:19 AM, Nick Clifton <nickc@redhat.com> wrote:
> Hi Guys,
>
>  I am applying the patch below to fix the assignment of adjacent ELF
>  sections to a segment.  At the moment the code looks like this:
>
>    else if (BFD_ALIGN (last_hdr->lma + last_size, maxpagesize)
>             < BFD_ALIGN (hdr->lma, maxpagesize))
>      {
>         /* If putting this section in this segment would force us to
>            skip a page in the segment, then we need a new segment.  */
>         new_segment = TRUE;
>      }
>
>  which does not work as intended because BFD_ALIGN always rounds its
>  argument up to the next alignment boundary.  So for example if one
>  section ended at 0x1fff and the next (potential) section to be
>  allocated started at 0x2001 and the pagesize was 0x1000 then the
>  comparison above would check "0x2000 < 0x3000" and so decide that a
>  new segment was needed.
>
>  The revised test looks like this:
>
>    else if (BFD_ALIGN (last_hdr->lma + last_size, maxpagesize) + maxpagesize
>            < hdr->lma)
>
>  so that it rounds up the end address of the previous section but
>  compares it directly with the start address of the current section.
>  The "+ maxpagesize" is there so that we actually allow for sections
>  that are in separate pages but not more than one page apart.  The
>  code has to be careful however to handle the case where rounding up
>  the end address of the previous section wraps around the end of the
>  address space so the actual patch looks like this:
>

Hi Nick,

Is there a testcase for this?

Thanks.

-- 
H.J.