Ok, I have enacted my changes and have created a
pretty decent patch for pth-1.4.1 which you can
get here:


http://www.geocities.com/doug_hilton/

I decided against posting it to the list as it was
around 8k when gzipped.

I have run minimal tests so far, but the pthread
test appears to be working.

To build it just do "./configure --enable-pthread"

I will send the diff in to GNU shortly.

Q: Should I make a patch for Debian's pth-1.4.0 ?

Cheers!
- Doug

Earlier, I wrote:

Reply-To: [EMAIL PROTECTED]

This, I imagine will be a bit of good news :-)

I had to do some severely ugly hacking to get
configure past the well-known sigaltstack problem,
but once I was able to do this, I was able to
compile pth using the following configure switches:

./configure --with-mctx-mth=sjlj --with-mctx-dsp=sjljlx \
 --with-mctx-stk=none --enable-pthread

What we are doing here is taking advantage of the fact
that pth provided for the 2.0 linux kernel which did
not have functioning sigstack or sigaltstack, to the
mctx-dsp=sjljlx was an i386 dependant hack around the
problem.

My hypothesis was the Hurd-i386 would be similar enough
that this would work. I was able to compile pth and
it passed its self-tests all OK.

Certainly further testing is in order, and I will attempt
to make a Hurd-friendly patch for pth and get it posted
asap.

- Doug








Reply via email to