PERFORCE change 118502 for review

[John Baldwin]

On Friday 20 April 2007 07:33:51 pm Matt Jacob wrote:
> http://perforce.freebsd.org/chv.cgi?CH=118502
> 
> Change 118502 by mjacob at mjexp on 2007/04/20 23:32:53
> 
> 	While it's true we have motherboard blacklists for MSI
> 	breakage, it's true that this is incomplete. Save ourselves
> 	some grief by not trying to do MSI for Ultra320 chips.
> 	
> 	MFP4 after:	1 days

Is it due to a broken motherboard or broken mpt(4) silicon?  If the former, do 
you have a system where it breaks?

> Affected files ...
> 
> .. //depot/projects/mjexp/sys/dev/mpt/mpt_pci.c#9 edit
> 
> Differences ...
> 
> ==== //depot/projects/mjexp/sys/dev/mpt/mpt_pci.c#9 (text+ko) ====
> 
> @@ -522,11 +522,13 @@
>  
>  	/* Get a handle to the interrupt */
>  	iqd = 0;
> +
>  	/*
>  	 * First try to alloc an MSI-X message.  If that
>  	 * fails, then try to alloc an MSI message instead.
> +	 * Don't do this for U320 chips.
>  	 */
> -	if (pci_msix_count(dev) == 1) {
> +	if (mpt->is_spi == 0 && pci_msix_count(dev) == 1) {
>  		mpt->pci_msi_count = 1;
>  		if (pci_alloc_msix(dev, &mpt->pci_msi_count) == 0) {
>  			iqd = 1;
> @@ -534,7 +536,7 @@
>  			mpt->pci_msi_count = 0;
>  		}
>  	}
> -	if (iqd == 0 && pci_msi_count(dev) == 1) {
> +	if (mpt->is_spi == 0 && iqd == 0 && pci_msi_count(dev) == 1) {
>  		mpt->pci_msi_count = 1;
>  		if (pci_alloc_msi(dev, &mpt->pci_msi_count) == 0) {
>  			iqd = 1;
> 



-- 
John Baldwin