Re: [PATCH] PCI: brcmstb: Declare a bitmap as a bitmap, not as a plain 'unsigned long'

[Krzysztof Wilczyński]

Hi Christophe!

[...]
> This bitmap can be BRCM_INT_PCI_MSI_LEGACY_NR or BRCM_INT_PCI_MSI_NR long.

Ahh.  OK.  Given this an option would be to: do nothing (keep current
status quo); allocate memory dynamically passing the "msi->nr" after it
has been set accordingly; use BRCM_INT_PCI_MSI_NR and waste a little bit
of space.

Perhaps moving to using the DECLARE_BITMAP() would be fine in this case
too, at least to match style of other drivers more closely.

Jim, Florian and Lorenzo - is this something that would be OK with you,
or you would rather keep things as they were?

> Addresses-Coverity: "Out-of-bounds access (ARRAY_VS_SINGLETON)"

This tag would have to be written as:

  Addresses-Coverity: ("Out-of-bounds access (ARRAY_VS_SINGLETON)")

[...]
> +	DECLARE_BITMAP		(used, BRCM_INT_PCI_MSI_NR);

Probably not the most elegant solution, but I would keep it as:

  DECLARE_BITMAP(used, BRCM_INT_PCI_MSI_NR);

Otherwise aligning either before or after the open bracket will cause
either an error or a warning issued by checkpatch.pl accordingly about
the style.  Other users of this (a vast majoirty) macro don't do any
specific alignment at large

[...]
> +	/*
> +	 * Sanity check to make sure that the 'used' bitmap in struct brcm_msi
> +	 * is large enough.
> +	 */
> +	BUILD_BUG_ON(BRCM_INT_PCI_MSI_LEGACY_NR > BRCM_INT_PCI_MSI_NR);

A healthy paranoia, I see. :-)

	Krzysztof