Subject: Re: DRI/DRM Intel 945GM problem
To: None <tech-x11@netbsd.org>
From: Ahmad M. Afuni <pseudo.statistic@gmail.com>
List: tech-x11
Date: 05/21/2007 22:05:13
Hey,
Same config here as well, actually.
Anyway, I have come to the conclusion that the lockup wasn't caused by
your patch, but by the -current kernel of today.
At the moment I'm stuck using a kernel I compiled on Saturday, 12th
May, since it seems that the current kernel is forcing me into the
debugger every time I try compiling something.
Anyway, I actually tried wip/modular-xorg-server-mesa653 to see if I
could get DRI working there, but same issue:
[drm] failed to load kernel module "i915"
(II) intel(0): [drm] drmOpen failed
(EE) intel(0): [dri] DRIScreenInit failed. Disabling DRI.
And I also had those keyboard problems as you mentioned, so I switched
back to x11/modular-xorg-server.
Furthermore, thinking maybe I had screwed up the i810 driver
compilation, I recompiled that, and am now using the 'intel' driver
from the Xorg FTP-- still no dri.
I'm probably still missing something.
Ahmad
On 5/21/07, Chavdar Ivanov <ci4ic4@gmail.com> wrote:
> On 20/05/07, Ahmad M. Afuni <pseudo.statistic@gmail.com> wrote:
> > Hey,
> >
> > Applied your patch but, unfortunately, it caused my system to lock up.
> > Xorg.0.log says the same thing about being unable to load the module.
>
> I've had the lock-ups if AIGLX is not disabled in this setup -
> attached is my xorg.conf, if it may be of any help.
>
> >
> > Thanks anyway,
> >
> > Ahmad
> > On 5/20/07, Chavdar Ivanov <ci4ic4@gmail.com> wrote:
> > > On 20/05/07, Ahmad M. Afuni <pseudo.statistic@gmail.com> wrote:
> > > > Hey,
> > > > I compiled my kernel (GENERIC.MP) with:
> > > > i915drm* at vga? # Intel i915, i945 DRM driver
> > > > And so dmesg tells me:
> > > > i915drm0 at vga1
> > > > info: [drm] Intel i945GM (unit 0)
> > > > info: [drm] AGP at 0xc0000000 256MB
> > > > info: [drm] Initialized i915 1.5.0 20060119
> > > > Which seems OK.
> > > > Now, to install xorg, I installed (from pkgsrc) x11/xf86-video-i810,
> > > > modular-xorg-server, and the usual mouse/keyboard input drivers.
> > > > I had:
> > > > X11_TYPE=modular
> > > > PKG_OPTIONS.xorg-server=dri
> > > > in my mk.conf.
> > > > Now, after installing the above, I tried X out, only to have this in
> > > > my Xorg.0.log:
> > > > drmOpenDevice: node name is /dev/dri/card0
> > > > drmOpenDevice: open result is 8, (OK)
> > > > drmOpenDevice: node name is /dev/dri/card0
> > > > drmOpenDevice: open result is 8, (OK)
> > > > [drm] failed to load kernel module "i915"
> > >
> > > I am sure I have been there at some point, but I can hardly remember
> > > exactly how I fixed it. I've kinda lost the number of
> > > modular-xorg-server* and mesa-dri* compilations; it's good fun,
> > > anyway.
> > >
> > > You got your /dev/dri/card0 as expected with the right mode?
> > >
> > > You followed generally what is in:
> > > http://mail-index.netbsd.org/tech-x11/2007/03/19/0000.html ?
> > >
> > > Maybe the following will help you (it did it for me), taken against
> > > -current from a few days ago:
> > >
> > > +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> > > --- sys/dev/pci/agp_i810.c.orig 2007-05-14 20:13:55.000000000 +0100
> > > +++ sys/dev/pci/agp_i810.c 2007-05-14 14:43:51.000000000 +0100
> > > @@ -71,28 +71,6 @@
> > > } \
> > > } while (0)
> > >
> > > -#define CHIP_I810 0 /* i810/i815 */
> > > -#define CHIP_I830 1 /* 830M/845G */
> > > -#define CHIP_I855 2 /* 852GM/855GM/865G */
> > > -#define CHIP_I915 3 /* 915G/915GM/945G/945GM */
> > > -
> > > -struct agp_i810_softc {
> > > - u_int32_t initial_aperture; /* aperture size at startup */
> > > - struct agp_gatt *gatt;
> > > - int chiptype; /* i810-like or i830 */
> > > - u_int32_t dcache_size; /* i810 only */
> > > - u_int32_t stolen; /* number of i830/845 gtt entries
> > > - for stolen memory */
> > > - bus_space_tag_t bst; /* register bus_space tag */
> > > - bus_space_handle_t bsh; /* register bus_space handle */
> > > - bus_space_tag_t gtt_bst; /* GTT bus_space tag */
> > > - bus_space_handle_t gtt_bsh; /* GTT bus_space handle */
> > > - struct pci_attach_args vga_pa;
> > > -
> > > - void *sc_powerhook;
> > > - struct pci_conf_state sc_pciconf;
> > > -};
> > > -
> > > static u_int32_t agp_i810_get_aperture(struct agp_softc *);
> > > static int agp_i810_set_aperture(struct agp_softc *, u_int32_t);
> > > static int agp_i810_bind_page(struct agp_softc *, off_t, bus_addr_t);
> > > @@ -725,7 +703,7 @@
> > > * Until the issue is solved, simply restore it.
> > > */
> > >
> > > -#if 0
> > > +#if 1
> > > regval = bus_space_read_4(isc->bst, isc->bsh, AGP_I810_PGTBL_CTL);
> > > if (regval != (isc->gatt->ag_physical | 1)) {
> > > printf("agp_i810_bind_memory: PGTBL_CTL is 0x%x - fixing\n",
> > > --- sys/dev/pci/agpvar.h.orig 2007-05-14 20:13:55.000000000 +0100
> > > +++ sys/dev/pci/agpvar.h 2007-05-14 14:45:43.000000000 +0100
> > > @@ -161,6 +161,28 @@
> > > size_t ag_size;
> > > };
> > >
> > > +#define CHIP_I810 0 /* i810/i815 */
> > > +#define CHIP_I830 1 /* 830M/845G */
> > > +#define CHIP_I855 2 /* 852GM/855GM/865G */
> > > +#define CHIP_I915 3 /* 915G/915GM/945G/945GM */
> > > +
> > > +struct agp_i810_softc {
> > > + u_int32_t initial_aperture; /* aperture size at startup */
> > > + struct agp_gatt *gatt;
> > > + int chiptype; /* i810-like or i830 */
> > > + u_int32_t dcache_size; /* i810 only */
> > > + u_int32_t stolen; /* number of i830/845 gtt entries
> > > + for stolen memory */
> > > + bus_space_tag_t bst; /* register bus_space tag */
> > > + bus_space_handle_t bsh; /* register bus_space handle */
> > > + bus_space_tag_t gtt_bst; /* GTT bus_space tag */
> > > + bus_space_handle_t gtt_bsh; /* GTT bus_space handle */
> > > + struct pci_attach_args vga_pa;
> > > +
> > > + void *sc_powerhook;
> > > + struct pci_conf_state sc_pciconf;
> > > +};
> > > +
> > > int agpbusprint(void *, const char *);
> > >
> > > /*
> > > --- sys/dev/drm/drm_drv.c.orig 2007-05-14 20:14:08.000000000 +0100
> > > +++ sys/dev/drm/drm_drv.c 2007-05-14 14:49:54.000000000 +0100
> > > @@ -526,7 +526,7 @@
> > > }
> > >
> > > for(i = 0; i<DRM_MAX_PCI_RESOURCE; i++) {
> > > - if (dev->pci_map_data[i].mapped > 1) {
> > > + if (dev->pci_map_data[i].mapped > 0) {
> > > bus_space_unmap(dev->pci_map_data[i].maptype,
> > > dev->pci_map_data[i].bsh,
> > > dev->pci_map_data[i].size);
> > > @@ -691,7 +691,7 @@
> > > i = 0;
> > >
> > > for (i = 0; i < DRM_MAX_PCI_RESOURCE; i++)
> > > - if (dev->pci_map_data[i].mapped != 0)
> > > + if (dev->pci_map_data[i].mapped > 0)
> > > {
> > > bus_space_unmap(dev->pa.pa_memt,
> > > dev->pci_map_data[i].bsh,
> > > --- sys/dev/drm/drm_memory.c.orig 2007-05-14 20:14:08.000000000 +0100
> > > +++ sys/dev/drm/drm_memory.c 2007-05-14 14:51:39.000000000 +0100
> > > @@ -97,8 +97,9 @@
> > > map->bst = dev->pa.pa_memt;
> > > map->cnt = &(dev->pci_map_data[i].mapped);
> > > map->mapsize = dev->pci_map_data[i].size;
> > > - dev->pci_map_data[i].mapped++;
> > > - if (dev->pci_map_data[i].mapped > 1)
> > > + if (dev->pci_map_data[i].mapped >= 0)
> > > + dev->pci_map_data[i].mapped++;
> > > + if (dev->pci_map_data[i].mapped != 1)
> > > {
> > > map->bsh = dev->pci_map_data[i].bsh;
> > > return dev->pci_map_data[i].vaddr;
> > > --- sys/dev/pci/drm/i915_drv.c.orig 2007-05-14 20:13:55.000000000 +0100
> > > +++ sys/dev/pci/drm/i915_drv.c 2007-05-14 14:54:36.000000000 +0100
> > > @@ -39,6 +39,8 @@
> > > #include "i915_drm.h"
> > > #include "i915_drv.h"
> > > #include "drm_pciids.h"
> > > +#include "agp_i810.h"
> > > +
> > >
> > > /* drv_PCI_IDs comes from drm_pciids.h, generated from drm_pciids.txt. */
> > > static drm_pci_id_list_t i915_pciidlist[] = {
> > > @@ -134,9 +136,34 @@
> > > struct pci_attach_args *pa = aux;
> > > drm_device_t *dev = (drm_device_t *)self;
> > >
> > > +#if NAGP_I810 > 0
> > > + struct agp_softc *agpsc;
> > > + struct agp_i810_softc *agpi810sc;
> > > + int reg;
> > > +#endif
> > > +
> > > +
> > > i915_configure(dev);
> > >
> > > drm_attach(self, pa, i915_pciidlist);
> > > +
> > > +#if NAGP_I810 > 0
> > > + agpsc = (struct agp_softc *)agp_find_device(0);
> > > + if(agpsc == NULL || agpsc->as_chipc == NULL)
> > > + return;
> > > +
> > > + agpi810sc = agpsc->as_chipc;
> > > + if(agpi810sc->chiptype == CHIP_I915)
> > > + reg = 0;
> > > + else
> > > + reg = 1;
> > > +
> > > + dev->pci_map_data[reg].bsh = agpi810sc->bsh;
> > > + dev->pci_map_data[reg].mapped = -1; /* never unmap */
> > > + dev->pci_map_data[reg].vaddr =
> > > + bus_space_vaddr(agpi810sc->bst, agpi810sc->bsh);
> > > +#endif
> > > +
> > > }
> > >
> > > CFATTACH_DECL(i915drm, sizeof(drm_device_t), i915drm_probe, i915drm_attach,
> > > --------------------------------------------------------
> > >
> > > > (II) I810(0): [drm] drmOpen failed
> > > > (EE) I810(0): [dri] DRIScreenInit failed. Disabling DRI.
> > > > Afterwards, realizing that I needed mesa-dri, I installed wip/mesa-dri
> > > > and copied over the MesaLib.buildlink3.mk file over to
> > > > graphics/MesaLib/buildlink3.mk.
> > >
> > > I am presently running Xorg7.2+Mesa652+dri on my laptop; the only
> > > problem is I could not get AIGLX to work. Of the more interesting
> > > OpenGL applications available, I got Blender to run accelerated, and
> > > of course BZFlag!
> > >
> > > I got even Xorg7.3 + Mesa653+dri working, but for some reason the
> > > server did not recognize the 'vt05' option, which lead to some weird
> > > effects when used with gdm - like sometimes the keyboard attached to
> > > one of the wscons terminals behind... also on occasions some keys got
> > > repeated many times with no reason; this was even running with AIGLX
> > > and Composite and I was able to try starting Compiz and Beryl; the
> > > latter kinda ran, but gave mostly white background with occasional
> > > window brought in with ALT-Tab; the former did tjhe '--replace' bit
> > > but then refused to continue due to the lack of a particular call.
> > > I've since reverted back to the initial accelerated setup - mainly
> > > because of the problem with the VTs (and also because otherwise I had
> > > to recompile every OpenGL application - that's another story).
> > >
> > > > Furthermore, I pkg_deleted i810's driver and rebuilt it.
> > >
> > > I also did that just in case, although I don't believe it made any difference.
> > >
> > > > Still, exact same error. I must've missed something.
> > > > Anyone know what that might be?
> > > >
> > > > Thanks,
> > > > Ahmad
> > > >
> > >
> > > Chavdar Ivanov
> > >
> >
>
>