On Tue, Dec 28, 1999 at 05:19:16PM -0800, Jason Evans wrote:
> It probably isn't currently reasonable to do cancellation points correctly,
> partially correctly, or even partially, without causing the deadlocks that
> you point out. I knew that the cancellation points were extended into
> libc, but it didn't occur to me that it would cause the deadlocks. I'm
> inclined to completely remove the cancellation points from the linuxthreads
> port until we have better support available in libc. Any objections?
There is no good solution. But, removing the libc cancellation points is,
IMO, the best of a set of bad choices. This is what the older
linuxthreads port did.
> It's clear that we need to add additional entry points to libc in order to
> support cancellation points, as well as to allow creation of a real
> libpthread, rather than libc_r. This is getting close to the top of my
> priority list now, but my experience in such things is limited, so it will
> be quite a learning experience. In other words, it could take me a long
> while, and if any more knowledgeable person would like to take this job, I
> wouldn't complain. =)
I'm not volunteering. But, I promised you some thoughts on how to do it,
which I still owe you. I don't think its "hard", just tedious.
--
Richard Seaman, Jr. email: [EMAIL PROTECTED]
5182 N. Maple Lane phone: 262-367-5450
Chenequa WI 53058 fax: 262-367-5852
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message