Duncan Webb wrote:
James Gatt wrote:
I thought I should ask the obvious question before delving into
details - has anybody tried this configuration:
Gentoo Linux
mplayer (1.0pre5 - not the latest but I have other issues with later
versions.)
amd64
DirectFB-0.9.24
Matrox G450
I'm surprised that you could find a AMD64 motherboard that has a AGP
bus, most seem to have PCI-E busses.
There are a few S939 <-> AGP chipsets around. It's just as well really
as I don't know of any PCI-E graphics cards which do analogue TVOut as
well as the Matrox ones.
First run ldd
/usr/lib64/directfb-0.9.24/gfxdrivers/libdirectfb_matrox.so, if you
have not found errors then run "revdep-rebuild -p" (in gentoolkit)
# ldd /usr/lib64/directfb-0.9.24/gfxdrivers/libdirectfb_matrox.so
libz.so.1 => /lib/libz.so.1 (0x00002aaaaabd3000)
libm.so.6 => /lib/tls/libm.so.6 (0x00002aaaaace9000)
libdl.so.2 => /lib/libdl.so.2 (0x00002aaaaae6f000)
libpthread.so.0 => /lib/tls/libpthread.so.0 (0x00002aaaaaf72000)
libc.so.6 => /lib/tls/libc.so.6 (0x00002aaaab088000)
/lib64/ld-linux-x86-64.so.2 (0x0000555555554000)
This looks ok to me. (I'm still not used to these long 64-bit addresses
but I'm getting there...)
Running revdep-rebuild (I had already done this) complains about
dfbshow, dfbswitch, lib++dfb-0.9.so.22.0.0 and XDirectFB. This is
expected as I have not rebuilt these since upgrading to DirectFB-0.9.24,
but I don't believe any of these are used by the mplayer/DirectFB
combination.
Secondly try adding disable-module=lirc in ~/.directfbrc, the lircd
conflicted with directfb exiting.
I really don't want to have to do that, as I am using lirc input to
DirectFB in my own app. (This works perfectly in 0.9.22 and I have had
no exit problems there with mplayer or my own app.) Not having lirc
would block me from using 0.9.24.
I tried it anyway, and can confirm that it did something as I got in the
output:
(*) Direct/Modules: suppress module 'lirc'
However, this did not affect the crash.
Thirdly what modules have you got loaded?
Not many:
# lsmod
Module Size Used by
lirc_serial 12512 1
8139too 21632 0
I have (and always have had) fusion compiled directly into the kernel,
so this won't show here.
I don't think I have shed much light on the problem here. I rather
suspect that it's something that is not 64-bit clean in maven_init() or
the code calling it. This doesn't completely surprise me as I suspect
there are not many people putting G450 cards into amd64 machines, so
this code won't be well tested in the field. I am happy to do some of
this testing, but I don't know the directfb matrox driver code (yet?).
Regards,
James.
_______________________________________________
directfb-users mailing list
[email protected]
http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-users
