Grr, I've fixed that at least once.  You are correct, this is wrong.  
Others have IPMI this on ia64, and it does work.

The following patch fixes the problem (changes it to (IPMI_TIMEOUT_TIME 
* HZ) / 1000).  It also contains fixes for the previous patch I send 
that you had problems with.

Linus, assuming Rusty doesn't have any problems with this, could you apply?

Thanks for pointing this out,

-Corey

Randy.Dunlap wrote:

>Hi Corey,
>
>Is IPMI meant to work on ia64?  I'm not trying to use it,
>just build it.  I'm getting gcc warnings in impi_msghandler.c,
>whenever IPMI_TIMEOUT_JIFFIES is referenced.  The warning message
>is "division by zero" since HZ is 1024 and the #define is
>(IPMI_TIMEOUT_TIME/(1000/HZ))
>
>Is there an assumption here that HZ is 1000?  If so, that's not
>good.
>
>--
>~Randy   [on 2.6.0-test4]
>
>
>  
>