Re: [Dri-devel] Assertion failure in savageBindTexture (was: Testing a theory about Savage lockups)

[Felix <fx...@gm...>]

On Mon, 8 Mar 2004 00:05:37 +0100
Felix K=FChling <fx...@gm...> wrote:

> On Sun, 7 Mar 2004 17:35:45 -0300
> "Paulo R. Dallan" <pr...@uo...> wrote:
>=20
> > Em Qui 04 Mar 2004 19:20, Felix K=FChling escreveu:
> [snip]
> >=20
> > Hello Felix,
> >=20
> > Installed the new tree; Haven't tested it fully with wolf3d-et yet, but=
 in=20
> > Vegastrike, and, at a certain moment, had a similar error, already with=
 the=20
> > new driver:
> >=20
> > 2.Loading bmp file cockpit/selectbox.png alp cockpit/selectbox.png 2.Lo=
ading=20
> > bmp                                             file cockpit/selectbox.=
png=20
> > alp cockpit/selectbox.png 2.Loading bmp file=20
> > cockpit                                            /selectbox.png alp=20
> > cockpit/selectbox.png 2.Loading bmp file cockpit/selectbox.pn          =
                                 =20
> > g alp cockpit/selectbox.png 2.Loading bmp file cockpit/selectbox.png al=
p=20
> > cockpit                                            /selectbox.png vegas=
trike:=20
> > savagetex.c:1837: savageBindTexture: Assertion `(targ                  =
                         =20
> > et !=3D 0x0DE1) || (tObj->DriverData !=3D ((void *)0))' failed.
> >=20
> > So I believe it did not work yet! :(
>=20
> Right. I saw it too with Flight Gear. It seems to happen when the driver
> runs out of texture memory. There also seems to be a texture memory
> leak. Therefore it's more likely to happen after a long-running session.
> I'll see if I can track it down.

Done. At least it should work with one context now. I'm considering to
port the texture memory management to Ian's texmem code as it seems the
savage driver is currently duplicating lots of this code (modulo bugs
;-).

>=20
[snip]

Felix