So here's a problematic portion:
84->io_write_request ("/* tc-mips.c -- assemble code for a MIPS chip.
Copyright 1993, 1994, 1995, 19" -1)task878-> 3206 (pn{ 1}) = 0
There should have been a "> RPT ..." notation here, I think. (I don't
remember rpctrace output details, so I'm just reconstructing what to expect
from context.) This request didn't get a reply.
84->interrupt_operation () = 0
Here's libc interrupting it.
task878-> 3206 (pn{ 1}) = 0
task878-> 3206 (pn{ 1}) = 0
task878-> 3206 (pn{ 1}) = 0
task878-> 3205 (pn{ 29}) = 0
task878-> 3206 (pn{ 1}) = 0
task878-> 3205 (pn{ 34}) = 0
task878-> 3206 (pn{ 1}) = 0
task878-> 3208 (pn{ 2} 1 -1) = 0
84->io_write_request ("/* tc-mips.c -- assemble code for a MIPS chip.
Copyright 1993, 1994, 1995, 19" -1) > 81 ...
And here's libc restarting the RPC. But it appears not to have waited for
the reply after it sent interrupt_operation.
reply? 102 > 0 463659
And here's the reply to the first io_write request. libc was supposed
to wait for this reply after it sent interrupt_operation and before it
restarted the RPC.
reply? 81 > 0 463659
And here's the reply to the second io_write, so it's done it twice.
