Comment 19 for bug 214442

I'm afraid the patch is not acceptable in this form. If anything, it should default to the span locking being enabled, with the possibility of disabling it via an environment variable, as is done in the r300 driver. However, AFAIK the latter hasn't actually needed it in a while (certainly not in my experience), so it would be even better to identify and fix the actual cause of the hangs. It sounds like other parts of the driver may not handle the locking correctly.