Tiago:
Yes, your patch is right. i have replaced the movdqa at
imlib2/src/lib/amd64_blend.S line 1267 with movdqu and it worked fine. the
original source was...
SIZE(imlib_amd64_copy_rgba_to_rgba)
PR_(imlib_amd64_copy_rgb_to_rgba):
ENTER
movdqa mX000X000X000X000(%rip), %xmm5
leaq (%rsi, %r8, 4), %rsi
leaq (%rdi, %r8, 4), %rdi
and resulting source was:
SIZE(imlib_amd64_copy_rgba_to_rgba)
PR_(imlib_amd64_copy_rgb_to_rgba):
ENTER
movdqu mX000X000X000X000(%rip), %xmm5
leaq (%rsi, %r8, 4), %rsi
leaq (%rdi, %r8, 4), %rdi
It solved my problem and now emblem works fine. Thanks a lot!
Ivan Hernandez (zaikxtox)
El Dom 06 Nov 2005 21:10, Tiago Victor Gehring escribió:
> I had the same problem on my amd64 computer, the problem is the
> (assembly) amd64 optimized routines in imlib2; It's just a matter of
> replacing a couple of movdqa instructions (aligned mov) to movdqu
> (unaligned mov), there's practically no performance hit (I guess)
> because this instructions are outside the loops that actually copy the
> data;
> This has already been discussed and one patch for the case of the
> function ..._copy_rgba_to_rgba even committed (this was the one that
> originally gave me problems); I had a small patch for the other cases
> but have to re-create it because my amd64 notebook is being repaired and
> I'm only with a Winblows computer for work.. :-)
> Anyway, I'll try to do it (I mean re-create the patch) tomorrow ok,
> so you can test it and see if it works..
>
> Tiago Gehring
>
> [EMAIL PROTECTED] wrote:
> > I have yesterdey get a clean cvs version after trying the last e source
> > on my office. On i686 it works fine, and a lot of changes and new thing
> > were happening from the last update i did to now... but curiously, on
> > amd64, i found myself unable to run emblem. It's not a pain for me
> > because i can set up the background with enlightenment_remote, but doing
> >
> > a little more os investigation with gdb it says:
> > ::[EMAIL PROTECTED] (08:29:45) - ~/Documentos/src/e17 }::
> >
> > .oOo. gdb -e emblem
> > GNU gdb 6.3-debian
> > This GDB was configured as "x86_64-linux".
> > (gdb) run
> > Starting program: /usr/local/bin/emblem
> > Using host libthread_db library "/lib/libthread_db.so.1".
> > [Thread debugging using libthread_db enabled]
> > [New Thread 182928639344 (LWP 15810)]
> > Unable to load file,
> >
> > Program received signal SIGSEGV, Segmentation fault.
> > [Switching to Thread 182928639344 (LWP 15810)]
> > 0x0000002a965c13fb in __imlib_amd64_copy_rgb_to_rgba ()
> > from /usr/local/lib/libImlib2.so.1
> > (gdb)
> >
> > i have imported the code as anjuta project (the only one ide i know how
> > to debug correctly) and apparently the "Unable to load file," has
> > nothing to do with the bug. it comes a lot earlier.
> >
> > Also, i have dumped my enlightenment conf dir (mv .e .e.notusing) and it
> > continued this way.
> > I have downloaded again imlib2 from cvs, on a clean directory and
> > compiled all again, libs, and e, and e_utils... but the bug is still
> > there.
> >
> > Asking on #e @ freenode.net i have found that someone else has the same
> > problem, on amd64, with another distribution.
> >
> > If someone knows how to get more information with gdb, and how to attach
> > the source to de debugger so i can give more info it will be nice.
> >
> > Thanks for all
> > Ivan Henrnandez (zaikxtox)
> >
-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. Download
it for free - -and be entered to win a 42" plasma tv or your very own
Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php
_______________________________________________
enlightenment-devel mailing list
enlightenment-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
