We have a user who is doing socket level readv between two OSF/1 v3.0 systems,
reading buffers ranging in size from 100K to 9MB. If he introduces brief
pauses between reads, by examining buffer before next read, the code works,
but if he does continuous reads, the system crashes with errors:
panic (cpu 0) : m_copym length
or m_copym offset
or receive 1
The alphas in question are connected via fddi. This user reports that the
same code works between suns running SunOS, connected via ethernet, or
between rs6000s connected via ethernet, and believes there is a bug in OSF/1
causing this problem. Does anyone have any insights, clues, or ideas?
Many thanks!
Judith Reed
judith_at_npac.syr.edu
Northeast Parallel Architectures Center
Syracuse University
Received on Wed Feb 22 1995 - 09:01:25 NZDT