Fw: Re: [ANNOUNCE] xf86-video-ati 6.9.0.91 - release candidate

[Alex Deucher]

On Fri, Dec 26, 2008 at 8:10 PM, Gene Heskett <gene.heskett at verizon.net> wrote:
> On Friday 26 December 2008, Alex Deucher wrote:
>>On Tue, Dec 23, 2008 at 11:57 PM, Shunichi Fuji <palglowr at gmail.com> wrote:
>>> Begin forwarded message:
>>>
>>> Date: Tue, 23 Dec 2008 22:29:38 -0500
>>> From: Gene Heskett <gene.heskett at verizon.net>
>>> To: Shunichi Fuji <palglowr at gmail.com>
>>> Subject: Re: [ANNOUNCE] xf86-video-ati 6.9.0.91 - release candidate
>>>
>>> On Tuesday 23 December 2008, Shunichi Fuji wrote:
>>>>On Tue, 23 Dec 2008 19:13:58 -0500
>>>>
>>>>Gene Heskett <gene.heskett at verizon.net> wrote:
>>>>> gcc -DHAVE_CONFIG_H -I. -I.. -I./AtomBios/includes -Wall
>>>>> -I/usr/include/xorg -I/usr/include/drm -I/usr/include/X11/dri
>>>>> -DDISABLE_EASF -DENABLE_ALL_SERVICE_FUNCTIONS -DATOM_BIOS
>>>>> -DATOM_BIOS_PARSER -DDRIVER_PARSER -g -O2 -MT radeon_accel.lo -MD -MP
>>>>> -MF .deps/radeon_accel.Tpo -c
>>>>> radeon_accel.c  -fPIC -DPIC -o .libs/radeon_accel.o
>>>>> radeon_accel.c: In function 'RADEONEngineInit':
>>>>> radeon_accel.c:383: error: 'RADEON_PARAM_NUM_GB_PIPES' undeclared (first
>>>>> use in this function)
>>>>> radeon_accel.c:383: error: (Each undeclared identifier is reported only
>>>>> once radeon_accel.c:383: error: for each function it appears in.)
>>>>> make[2]: *** [radeon_accel.lo] Error 1
>>>>> make[2]: Leaving directory `/usr/src/xf86-video-ati-6.9.0.91/src'
>>>>> make[1]: *** [all-recursive] Error 1
>>>>> make[1]: Leaving directory `/usr/src/xf86-video-ati-6.9.0.91'
>>>>> make: *** [all] Error 2
>>>>>
>>>>> What header am I missing?
>>>>
>>>>RADEON_PARAM_NUM_GB_PIPES defined in radon_drm.h, and git repository has
>>>> it one. but tar package seems don't have it.
>>>>
>>>>i guess that radeon_drm.h will ship with libdrm
>>>
>>> Not installed on this FU8 system, and not available to yumex.
>>>
>>>>and it install to
>>>> include/drm. there are 3 copy at least kernel,drm and xf86-video-ati...
>>>
>>> I found the one in xf86-video-radeonhd-1.2.4, and copied it to the src
>>> tree beside the other .h files, and it built, and I've installed it and
>>> running it, although not much is changed.  glxgears is still moping along
>>> at about 875 fps, and glxinfo says no drm?dri.
>>>
>>> But now this exposes a new problem I don't recall in these exact words,
>>> when I do a startx, it cannot load the kernel radeon module
>>> (this from the Xorg.0.log)
>>>
>>> (WW) RADEON(0): Direct rendering for RN50/RS600/R600 forced on -- This is
>>> NOT officially supported at the hardware level and may cause instability
>>> or lockups
>>> drmOpenDevice: node name is /dev/dri/card0
>>> drmOpenDevice: open result is -1, (No such device)
>>> drmOpenDevice: open result is -1, (No such device)
>>> drmOpenDevice: Open failed
>>> [drm] failed to load kernel module "radeon"
>>> (EE) RADEON(0): [dri] RADEONDRIGetVersion failed to open the DRM
>>> [dri] Disabling DRI.
>>>
>>> So, just for giggles:
>>>
>>> [root at coyote ~]# modprobe radeon
>>> WARNING: Error inserting drm
>>> (/lib/modules/2.6.28-rc9/kernel/drivers/gpu/drm/drm.ko): Invalid argument
>>> FATAL: Error inserting radeon
>>> (/lib/modules/2.6.28-rc9/kernel/drivers/gpu/drm/radeon/radeon.ko): Invalid
>>> argument
>>>
>>> I can also recall not too far back in the 2.6.28-rc cycle, seeing dmesg
>>> reporting that the firmware for the card had been loaded from
>>> the /lib/firmware/radeon directory, but that appears to be on the missing
>>> list too for an rcN or 2.  The firmware is there from previous kernel
>>> builds, but is no longer being installed by this line of my "./makeit"
>>> script:
>>>
>>> make firmware_install && \
>>>
>>> Which tells me something has changed in the Makefiles or .config's.
>>>
>>> How important is this firmware for the health & well being of an
>>> HD2400-Pro (rv610) card?
>>
>>There is no working drm support yet for r6xx/r7xx cards yet.  It will
>>be released soon.
>>
>>> From the .config that built this kernel:
>>>
>>> [root at coyote linux-2.6.28-rc9]# grep FIRMWARE .config
>>> # CONFIG_PREVENT_FIRMWARE_BUILD is not set
>>> # CONFIG_FIRMWARE_IN_KERNEL is not set
>>> CONFIG_EXTRA_FIRMWARE="radeon/R100_cp.bin.ihex  radeon/R200_cp.bin.ihex
>>> radeon/R300_cp.bin.ihex  radeon/R420_cp.bin.ihex  radeon/R520_cp.bin.ihex
>>> radeon/RS600_cp.bin.ihex  radeon/RS690_cp.bin.ihex"
>>> CONFIG_EXTRA_FIRMWARE_DIR="firmware"
>>> # CONFIG_DVB_AV7110_FIRMWARE is not set
>>> # CONFIG_FIRMWARE_EDID is not set
>>> CONFIG_FIRMWARE_MEMMAP=y
>>
>>Looks like you are using some distro-specific patch that separates the
>>firmware from the radeon drm.
>>
> I can nuke the stuff in /lib/firmware/radeon, and nuke that from the .config
> too I guess.
>
> The 2.6.28 final kernel does not install those anymore.  And I no longer see
> messages in the boot log indicating it is being loaded either.  I guess the
> answer to "Should I?" is yes?
>

for r1xx-r5xx and rs690 you should see something like this in dmesg:
[   39.963571] [drm] Loading RS690/RS740 Microcode

X loads the drm generally so you won't see anything until X starts.
You won't see anything at all if you have an r6xx or r7xx radeon since
they aren't yet supported in the drm.

> Time frame for the release?

Soon.  Just trying to tie up a few loose ends.

Alex