From: Jeff Dike <jdike@ka...> - 2002-09-06 16:17:07
> Commenting out a BUG_ON works but seems a bit drastic I think that is
> to insure that irq's are enabled when setup for the common buffer
> might want to start I/O.
The 2.5.33 patch is available, I just haven't announced it yet.
The IRQ thing is caused by the fact that I inverted the save_flags flags.
I set a bit for each enabled signal, while every other Linux in the world
sets a bit for each blocked signal. So, I inverted things in signal_kern.c.
Have a look at the s/enable/disable changes there if you're interested.
From: Jeff Dike <jdike@ka...> - 2002-09-06 18:07:34
> Sorry about the list not being in diff format (I am in the middle of
> switching source trees) I can do that later today if needed.
Don't bother. The UML 2.5.33 patch is on the mirrors.
Those changes look right, except for the irq one, which requires a little