On Nov 25, 2007 6:18 PM, Lionel Flandrin <simias.n_at_gmail.com> wrote: > "Vladimir Timofeev" <vovkasm_at_gmail.com> writes: > > > Hi, if u have nforce2 chipset on motheboard side, > > I do > > > try this patch... > > with this I can run X, but not test more... > > > > How to patch: > > cd /usr/src > > patch < latest-agp_nvidia.c.diff > > > > my conf: Abit NF7 motheboard, Radeon 9200 video, fbsd 7.0-BETA3, X > > Server 1.4.0 (from /var/log/Xorg.0.log) > > > > I have a zenith chaintech motherboard and a radeon 9600 pro, same > fbsd/X version, and indeed, it works, thanks a lot! I can finally use my > two screens again! > > I played with xrandr, glxgears and some 3d apps (blender for instance) > and it seems to work perfectly. Will this patch be commited to the main > tree ? May be, if this works and doing right things ;-) After some code cleanup and revision... I'am not commiter anything... may be Eric Anholt? > > Also, I'm curious, what was the problem in the first place ? I must > say I have no clue what this patch is doing (excepted it doesn't seem > too nasty :). Story of this patch is short. I take this kernel crash after upgrade to current (now releng 7). May be gcc4 change 'volatile' behaviour or anything else... I start studing code differences in linux and fbsd agp drivers code. In this patch for agp_nvidia_flush_tlb I realize two things: 1. get out volatile temporary variable to switch off potentialy compiler optimizations (in linux code this variable absent) 2. ensure that we read no more than allocated entries in agp gatt structure (in linux and current realization this read 33(magic?) entries) But it only my guesses. (I'am not c programmer, only know syntax ;-) > > Thanks again, > -- > Lionel >Received on Sun Nov 25 2007 - 16:24:02 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:23 UTC