« first day    last day (20 days later) » 

4:01 AM
Sorry, I saw your question just now.
After relogining into the server, the segmentation fault of /tmp/ls disappears
this is the mount options in /etc/fstab : /dev/sda5 /tmp ext2 defaults 0 1
The error message about gcc-4.8.2 still appears
$ echo "int main(void){ return 0; }" > ./ffconf.rgbbriKe.c && gcc -I/home/vis/guangli/local/include -c -o ./ffconf.LLalSg6X.o ./ffconf.rgbbriKe.c && gcc -g -L/home/vis/guangli/local/lib -ldl -o ./ffconf.G3SYKa9M ./ffconf.LLalSg6X.o
[vis@yq01-idl-gpu-offline104: lab] ->

$ ./ffconf.G3SYKa9M
[vis@yq01-idl-gpu-offline104: lab] ->

$ cp ff
ffconf.G3SYKa9M    ffconf.LLalSg6X.o  ffconf.rgbbriKe.c
[vis@yq01-idl-gpu-offline104: lab] ->

$ cp ffconf.G3SYKa9M /tmp/
[vis@yq01-idl-gpu-offline104: lab] ->

$ cd /tmp/
The output of /tmp/ls is correct now
$ cp /bin/ls /tmp/
[vis@yq01-idl-gpu-offline104: tmp] ->

$ pwd
/tmp
[vis@yq01-idl-gpu-offline104: tmp] ->

$ ./ls
3da6400ace09e7c703b035059037600b								 ls
caffe.INFO											 openmpi-sessions-vis@yq01-idl-gpu-offline104_0
compute_image_mean.ERROR									 sh-thd-1496748316
compute_image_mean.FATAL									 ssh-DdaOl11835
compute_image_mean.INFO										 ssh-dTbZsw7871
compute_image_mean.WARNING									 ssh-HVKyk30657
compute_image_mean.yq01-idl-gpu-offline104.yq01.baidu.com.vis.log.ERROR.20170607-171659.30924	 ssh-IDPFY11894
 
 
12 hours later…
3:49 PM
@Beanocean doesn't that show ls running just fine when copied to /tmp? I thought it didn't before... is it intermittent?
 
4:00 PM
@Beanocean the path to libc and libdl there looks really suspicious!
@Beanocean readelf -d /tmp/ffconf.G3SYKa9M and readelf -d /opt/compiler/gcc-4.8.2/lib64/ld-linux-x86-64.so.2 would be interesting
 

« first day    last day (20 days later) »