stability problem gt 430 with latest drivers
After about not quite 3 days of running our embedded system under heavy test load, the I2C EDID channel stops working and the display no longer updates, although X11 goes into an infinite loop. The graphics thread in the application appears to be deadlocked in the nvidia driver, invoking glibc schedule. A soft reboot alone does not reset the machine. The machine has to be completely depowered to reset the error state.
The problem occurs with an embedded system with a gt 430. It has been found with nvidia driver versions 319.23, 319.32, and 325.15, compiled with linux kernel version 3.9.2. The linux kernel is 32 bit, and both PAE and non-PAE have the problem.
The problem does not occur with geforce 9500 gt cards, identical software components.
The problem does not occur with driver version 256.44, kernel 220.127.116.11, with gt430 card.
Once the error has occurred, it is possible to log in with ssh. X11 must be killed with -9 and will not restart. i2cdump will not work on the EDID channels. When X11 is killed, the "frozen" display becomes blank. The xorg log files generated upon restart attempt are attached.
Modifying BIOS's is not an attractive option, as the software must work with equipment that has already been widely deployed. The hardware is non-AGP.
Attached is a tar archive with the output of the nvidia-bug-report script, x11 log files for the failed restart attempt, and the output of various /proc files to cover the absence of lspci, lsusb, modinfo which are not present on the target embedded system. The toolchain operates in a cross fashion and files are simply copied into the target, so there is no installation script as such. A dead system is available for analysis,
Thanks in advance,
Re: stability problem gt 430 with latest drivers
Please ask your questions here.
|All times are GMT -5. The time now is 12:32 AM.|
Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2015, Jelsoft Enterprises Ltd.
Copyright ©1998 - 2014, nV News.