Re: Core dump (semi) update

[Alfred M. Szmidt]

I'm a bit confused about the padding that we add to the core note
segment in elfcore.c, it seems that it is 8 bytes long.  If that is
the case, then I really can't see any problems in the reading code,
here is a small bit from my debug session:

,----
| Breakpoint 3, elfcore_grok_lwpstatus (abfd=0x8278438, note=0x10175f0)
|     at ../../sources.redhat.com/bfd/elf.c:6628
| 6628      if (note->descsz != sizeof (lwpstat)
| (top-gdb) p/x *note
| $32 = {namesz = 0x5, descsz = 0x1e0, type = 0x10, namedata = 0x829289c, 
|   descdata = 0x82928a4, descpos = 0x1760}
`----

If the core segment is indeed 8 bytes long, then 0x1760 is the correct
position in the file where we should start reading of the (first)
lwpstatus note.

,----[ core.668, first lwpstatus note segment ]
| 00001750: e001 0000 1000 0000 434f 5245 0000 0000  ........CORE....
| 00001760: 0000 0000 0000 0000 0000 0000 0000 0000  ................
| 00001770: 0000 0000 0000 0000 1f00 0000 1f00 0000  ................
| 00001780: 1f00 0000 1f00 0000 f46f 0101 0898 1b01  .........o......
| 00001790: bc6f 0101 0200 0000 18da 1701 0898 1b01  .o..............
| 000017a0: 646f 0101 e7ff ffff 0000 0000 0000 0000  do..............
| 000017b0: c6e0 0401 1700 0000 c6e0 0401 1700 0000  ................
| 000017c0: c6e0 0401 0000 0000 0000 0000 0000 0000  ................
| [...snip...]
| 00001940: 0500 0000 e001 0000 1000 0000 434f 5245  ............CORE
| 00001950: 0000 0000 0000 0000 0100 0000 0000 0000  ...............
`----

It does seem that the stuff that gdb tries to read is correct, atleast
with regard to the offsets in the core file.

Cheers and thanks.