Subject: Re: FW: Debugging shared objects in GDB
To: None <>
From: der Mouse <mouse@Rodents.Montreal.QC.CA>
List: tech-kern
Date: 05/03/2001 18:14:09
>> Second try with a diffrent group, sorry if this is not really kernel
>> related.
> It is not kernel related.

I, however, am bringing up something that is, so I'll leave it here. :)

>> [...]
> I see this too - also that gdb can "lose control" of the process
> being debugged such that instead of a single step the process starts
> running, yet the next time through single step will work correctly.

I regularly see gdb's ptrace() calls fail (with ENOMEM or something
like it), when a retry succeeds.

It appears that when ptrace was changed to be layered on top of the
same process address space R/W code procfs is, someone neglected to
notice that the underlying routines could then fail.  ptrace needs to,
at the very least, retry.

I've been meaning to deal with it, but so far haven't gotten the time
and inclination together at the same time.

					der Mouse

		     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B