Mailinglist Archive: radeonhd (223 mails)

< Previous Next >
Re: [radeonhd] 0x791E:0x1043:0x8287: ATI Radeon 1250
  • From: "R.M. Thomas" <rmthomas@xxxxxxxxxxx>
  • Date: Mon, 02 Jun 2008 01:39:41 +0100
  • Message-id: <4843414D.8090405@xxxxxxxxxxx>
Dave's post was indeed helpful. But first I must clarify something which did not appear in my original message (Saturday). I attached there a copy of the shell script which I wrote to carry out the build. This script did not appear in the archive, and as a result an important piece of information was not visible to readers. At the top of the script I had put the comment:

#
# REMARK:
# It was necessary to remove file
# /usr/include/xorg/xf86_ansic.h
# in order to avoid a runtime error "undefined symbol xf86stat".
#

At the time I thought (on the basis of some googling) that xf86_ansic.h was an obsolete relic (awaiting de-ansification), and because its removal gave no error messages during the build and allowed the x-server to at least start I did not worry too much about discarding it. Bad move.

Having seen Dave's post I did some instrumenting of routine drmOpenDevice() in file xf86drm.c myself. I always got zero for the structure members st.st_dev and st.st_rdev and inconsistent values for dev. I conclude that there is something seriously wrong with the definition of dev_t in my build. I haven't tried to find where the definition is located; all I really know is that sizeof(dev_t) is 8.

Although it seems that removal of xf86_ansic.h wrecks the build, I have just as much trouble if I retain xf86_ansic.h. In the latter case the build again gives no error messages, but at runtime I get a black screen when starting X and messages in Xorg.0.log about not finding any fonts.

This is my first build using git and I guess I am trying to do too much at once.

Mike


--
To unsubscribe, e-mail: radeonhd+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: radeonhd+help@xxxxxxxxxxxx

< Previous Next >