Current-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: yet another MKPIE=yes problem on build.sh release



On Apr 25,  6:03pm, aniou%smutek.pl@localhost (Piotr Meyer) wrote:
-- Subject: Re: yet another MKPIE=yes problem on build.sh release

| On Sun, Apr 25, 2010 at 11:12:24AM -0400, Christos Zoulas wrote:
|  
| > How about adding && !defined(PRIMARY_LOAD_ADDRESS) in bsd.sys.mk?
| 
| It should work but I want mimic current behaviour: PIE_* flags
| are (re)setted in many places under src/sys/arch/i386/stand (directly
| or by .include <../Makefile.bootxx>) - and MKPIE=no is (IMO) acceptable 
| shortcut for this. 
| 
| Look at output from, for example: grep PIE_ /usr/src/sys/arch/i386/stand/*/*
| 
| In fact, looks like MKPIE=no should be set for all dirs under
| /usr/src/sys/arch/i386/stand - PRIMARY_LOAD_ADDRESS is valid
| for most locations, but not for /usr/src/sys/arch/i386/stand/mbr/*
| - where PIE_* flags also are cleared - and finally it may lead
| to inconsistency (some subdirectories excluded from MKPIE by
| PRIMARY_LOAD_ADDRESS, some by direct assignment in their make).

We should centralize controls like this rather than sprinkling them
all over the place. Other archs need this. An intermediate solution
is to put it in Makefile.booters or Makefile.inc, but I think it is
better in the long term to have a special variable to set -Ttext, and
then check on that.

christos


Home | Main Index | Thread Index | Old Index