Bug 207667 - ld produces incorrect alignment output in kernel
Summary: ld produces incorrect alignment output in kernel
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Fedora
Classification: Fedora
Component: binutils
Version: 6
Hardware: All
OS: Linux
medium
medium
Target Milestone: ---
Assignee: Jakub Jelinek
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2006-09-22 13:21 UTC by Neil Horman
Modified: 2007-11-30 22:11 UTC (History)
0 users

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2006-09-22 13:40:30 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)

Description Neil Horman 2006-09-22 13:21:06 UTC
Description of problem:
when linking the linux kernel, ld produces a PT_LOAD segment, which according to
the linker script should be 4 byte aligned, but the p_align value for the
section is 0x1000

Version-Release number of selected component (if applicable):
fedora kernel 2.6.18-1.2682.fc6kdump and later (possibly earlier versions as well)

How reproducible:
always

Steps to Reproduce:
1.readelf -e /boot/vmlinux-2.6.18-1.2682.fc6kdump

  
Actual results:
Program Headers:
  Type           Offset   VirtAddr   PhysAddr   FileSiz MemSiz  Flg Align
  LOAD           0x001000 0xc1000000 0x01000000 0x266820 0x266820 R E 0x1000
  LOAD           0x267880 0xc1266880 0x01266880 0x162804 0x375638 RWE 0x1000
  NOTE           0x000000 0x00000000 0x00000000 0x00000 0x00000 R   0x4

Note that the second LOAD segment shows and Alignment of 0x1000, but the
VirtAddr and Offset values for the segment are not on 4096 byte boundaries

Expected results:
According the linker script for this kernel the Align value should be 0x4

Additional info:

Comment 1 Jakub Jelinek 2006-09-22 13:40:30 UTC
That's just misunderstanding what p_align means.  It doesn't mean that
the segment has to start on an p_align aligned address, but that p_vaddr %
p_align == p_paddr % p_align == p_offset % p_align.
Look at any ELF shared library or binary, you'll see the same:
E.g. on ppc64, /lib64/libc.so.6 has:
  LOAD           0x000000 0x00000080761b0000 0x00000080761b0000 0x17b99f
0x17b99f R E 0x10000
  LOAD           0x17bda0 0x000000807633bda0 0x000000807633bda0 0x0127b0
0x016670 RW  0x10000
The 0x10000 alignment here just ensures that the library will be usable with
any page size up to 64K, it is e.g. not possible to adjust it in a way where
it couldn't be mmaped on 64K page kernel.

Comment 2 Jakub Jelinek 2006-09-22 13:44:26 UTC
Here is what gABI 4.1, chapter 5 says on this:
p_align As ``Program Loading'' describes in this chapter of the processor
        supplement, loadable process segments must have congruent
        values for p_vaddr and p_offset, modulo the page size. This
        member gives the value to which the segments are aligned in
        memory and in the file. Values 0 and 1 mean no alignment is
        required. Otherwise, p_align should be a positive, integral
        power of 2, and p_vaddr should equal p_offset, modulo
        p_align.



Note You need to log in before you can comment on or make changes to this bug.