Peter Naulls wrote:
Lee Noar wrote:
Peter Naulls wrote:
Lee Noar wrote:
I've checked my version of libglib2 in /home/riscos/env/lib and have
found that it does contain PIC code.
Is this a recent build?
A recent build, yes, but of version glib2.0_2.16.6. I've just tried to
rebuild with the autobuilder, but it failed to download the source
(last-failure attached).
Perhaps you want to try again, I just reran it it was ok. Sometimes
one of the mirrors in the A record isn't sync, and you see this. If
true, probably we need to notice the wget errors.
Ok, I've got it to build now. I had previously built it with the -d
option to preserve the source. The fact that the old build tree was
still present must have caused some confusion, and deleting it fixed the
problem.
Presumably you do have all of the changes I made recenty.
Yes, I'm up to date with SVN, but I'm still getting PIC objects in the
static archive. Specifically for the functions in glib/gnulib and
glib/libcharset. This occurs in a native Linux (Ubuntu) build also.
Do you not see this in the archive you have built? PIC code is easy to
spot as it has an instruction pattern as follows at the beginning of the
function:
ADD r7,pc,r7
LDMIA r8,{r7,r8}
LDR r8,[r8,#0]
LDR r7,[r8,r7,LSL#4]
Perhaps this is deliberate by the developers to make the build easier (I
don't know if it actually does) or maybe an oversight that hasn't shown
up on other platforms.
Can anyone else confirm if they are seeing PIC code in the libglib
static archive. Actually, assuming you're using a code viewer (like Zap)
in RISC OS, it's probably easier to build Vala and search the binary -
objects stored in archives aren't always word aligned and don't
disassemble correctly.
Lee.
_______________________________________________
GCCSDK mailing list [email protected]
Bugzilla: http://www.riscos.info/bugzilla/index.cgi
List Info: http://www.riscos.info/mailman/listinfo/gcc
Main Page: http://www.riscos.info/index.php/GCCSDK
