On Sunday 26 February 2006 23:31, Wolfgang Hoffmann wrote:

On Sunday 26 February 2006 19:13, Wolfgang Hoffmann wrote:

quoted

Ok, I did some reading and just started a git bisect. I didn't find hints
on how to bisect if I'm only interested in changes to sky2.[ch], so I'm
taking the full kernel tree and skip testing those bisect steps that
didn't change sky2.[ch].

Looking at Carl-Daniels 0.13a and Stephens patch against 0.15 in this
thread, I'll patch each bisect step such that sky2_poll() has

       sky2_write32(hw, STAT_CTRL, SC_STAT_CLR_IRQ);
       if (sky2_read8(hw, STAT_LEV_TIMER_CTRL) == TIM_START) {
               sky2_write8(hw, STAT_LEV_TIMER_CTRL, TIM_STOP);
               sky2_write8(hw, STAT_LEV_TIMER_CTRL, TIM_START);
        }

after exit_loop. Is that ok?

I'll report as soon as I have results.

Bisect done:

4d52b48b43d0d1d5959fa722ee0046e3542e5e1b is first bad commit
    [PATCH] sky2: support msi interrupt (revised)

Reverting this commit in git head seems to work, at least the driver builds
and loads. Is that sane?

I'm currently testing this (without any further modifications), let's see
if it hangs or not.

Ok, no hangs yet.

This version passed a test scenario only 0.13a has survived so far. I'll 
continue to use this to give it more testing tomorrow.

Looking at the reverted commit, I wonder if modprobing sky2 with disable_msi=1 
is equivalent to reverting the commit?