> Argh! It appears that recent insmod versions rearrange the section
> sequence for some reason ... I'm starting to think that using the
Hmmm. I have old insmod, actually:
mordor:~$ /sbin/insmod -V
insmod version 2.1.121
> linker script hack is a bad idea. Could you try the modified version
> of retrieve_monitor_pages below, which doesn't rely on any linker
> script at all?
It does not help. :-\
> Please apply this patch to the current CVS version, re-run ./configure
> and delete the host-linux.ld file (to make sure it isn't used).
> After running 'insmod -v -m freemware.o', check out the output
> produced by the printk, and make sure that every page containing
> any module code/data (according to the insmod map) is indeed
> among those mapped by this routine.
All seems to be fine, but it still reboots.
Dec 30 23:58:14 mordor kernel: freemware: vm c2822000..c2822fff -> page 0000172fDec 30
23:58:14 mordor kernel: freemware: vm c2823000..c2823fff -> page 0000172eDec 30
23:58:14 mordor kernel: freemware: vm c2824000..c2824fff -> page 0000172dDec 30
23:58:14 mordor kernel: freemware: vm c2825000..c2825fff -> page 0000172cDec 30
23:58:14 mordor kernel: freemware: vm c2826000..c2826fff -> page 00001777
Sections: Size Address Align
.this 0000004c c2822000 2**2
.text 00003b59 c282204c 2**2
.fixup 00000018 c2825ba5 2**0
.rodata 000007ee c2825bbd 2**0
__ex_table 00000010 c28263ac 2**2
.data 00000000 c28263bc 2**2
.kstrtab 0000000a c28263bc 2**0
.bss 000004c8 c28263c8 2**2
Symbols:
00000000 a emulation.c
00000000 a host-linux.c
00000000 a monitor.c
00000000 a fault.c
c2822000 d __this_module
[skip]
c2826804 D monitor_pages
Bye,
Oleg
