Re: Did the fix for recycled thread ids uncover another bug?

[Paul Gilliam <pgilliam@us.ibm.com>]

Michael,

I am having trouble posting to the mailing list.  Or maybe it's any outgoing 
mail I'm having problems with.  I hope you get this.

On Tuesday 15 June 2004 10:59, Paul Gilliam wrote:
> On Monday 14 June 2004 16:22, Daniel Jacobowitz wrote:
> > On Mon, Jun 14, 2004 at 04:17:47PM -0700, Paul Gilliam wrote:
> > > I will post a follow-up with the details if someone tells me this is
> > > the right mail-list.
> >
> > I already did...
>
> I have done some more homework and found that the problem is with 6.1 and
> seems to be fixed in 6.1.0.90_2004-06-15-cvs (I just tried it).  This is
> all on the PowerPC.
>
> So now I need to find a patch for 6.1 that will fix this.
>
> I have attached four files: the first is the test case, tbug.c; two are
> transcripts from gdb 6.1 (with some back-porting) for ppc-32 and for
> ppc-64; the last is a transcript from gdb 6.1.0.90_2004-06-15-cvs for
> ppc-32..
>
> 'tbug' starts two threads, each using the same thread function which just
> waits a few seconds and returns.  Set a break-point in 'tf' (the thread
> function) and run.  In the 64-bit case, everything works as expected.  In
> the 32 bit case, we get a message 'reading register pc (#64): No such
> process' and then we're hosed.
>
> Thanks for any help,
>
> -=# Paul Gilliam #=-