BUG: getting erranesous command timeouts

[Dan Williams]

On Fri, 2008-05-23 at 15:22 +0200, Holger Schurig wrote:
> When I move the test
> 
>     else if (priv->resp_len[priv->resp_idx])
>        shouldsleep = 0;        /* We have a command response *
> 
> before
> 
>     else if (priv->cur_cmd)
>         shouldsleep = 1;        /* Can't send a command; one already running */
> 
> the the problem vanishes. I hope this is the correct fix ...

No, that sounds academically right.  We should be checking for a
response to the current command that's in-progress before deciding that
we should sleep because there's a command in-progress.

Is this the source of your CF regression from earlier today?

Dan