RE: [PATCH 7/7] CCISS: run through Lindent

[Miller, Mike (OS Dev)]

> -----Original Message-----
> From: Helgaas, Bjorn 
> Sent: Thursday, June 15, 2006 10:43 AM
> To: Miller, Mike (OS Dev)
> Cc: ISS StorageDev; linux-kernel@xxxxxxxxxxxxxxx; Andrew Morton
> Subject: Re: [PATCH 7/7] CCISS: run through Lindent
> 
> On Thursday 15 June 2006 08:49, Miller, Mike (OS Dev) wrote
> > > -----Original Message-----
> > > From: Helgaas, Bjorn
> > > Sent: Wednesday, June 14, 2006 6:14 PM
> > > To: Miller, Mike (OS Dev)
> > > Cc: ISS StorageDev; linux-kernel@xxxxxxxxxxxxxxx; Andrew Morton
> > > Subject: [PATCH 7/7] CCISS: run through Lindent
> > > 
> > > cciss is full of inconsistent style ("for (" vs. "for(", 
> lines that 
> > > end with whitespace, lines beginning with a mix of spaces & tabs, 
> > > etc).
> > > 
> > > This patch changes only whitespace.
> > > 
> > > Signed-off-by: Bjorn Helgaas <bjorn.helgaas@xxxxxx>
> > 
> > I agree that we had lots of whitespace garbage and 
> inconsistent styles 
> > in the driver but I'm not sure I like all the indentation being 
> > removed from the product table. It makes it a bit harder to 
> read, IMO.
> 
> If you are OK with the patch other than the product table, 
> how about if you apply the patch as-is, and I post a 
> follow-on patch to fix the indentation?
> 
> I'm contemplating more than just a white-space change, and 
> it'd probably be better to keep the "indent" patch to be 
> white-space changes only.

Yes, please submit a follow-on for the indentation.

> 
> The cciss_pci_device_id[] table lists a bunch of subvendor 
> and subdevice IDs.  Is there any reason to check those 
> sub-IDs explicitly?  In other words, we currently list:
> 
>         {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISS,
>          0x0E11, 0x4070, 0, 0, 0},
> 
> Could we replace that with:
> 
>         {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISS,
>          PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0},
> 
> This would potentially make the driver claim additional devices.
> But do COMPAQ/CISS devices with sub-IDs other than the listed 
> ones really exist anyway?

Please note that we have several pci ids: 

         {PCI_VENDOR_ID_COMPAQ, PCI_DEVICE_ID_COMPAQ_CISS[ABCD],
          0x0E11, 0x4070, 0, 0, 0},

The pci id identifies the controller family such as 53xx, 64xx, etc. We
use the subsystem pci id to identify the particular controller like 641,
642, etc.

I had added this to the driver we release on hp.com:

       { PCI_VENDOR_ID_HP, PCI_ANY_ID, PCI_ANY_ID, PCI_ANY_ID,
                PCI_CLASS_STORAGE_RAID << 8, 0xffff << 8, 0},

to bind to Smart Array controllers for which I don't have an id. Seems
to work okay but then I need another method for displaying product name,
etc. There are also additional fields in the products struct on which we
rely. Things like number of s/g's and number of outstanding commands,
for instance. These are supposed to be in the firmware's config table
but we find they lie to us most of the time. :)

Lots of words, not much meaning.

mikem

> 
> Bjorn
> 
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/