Re: [ACPI] RFC: ACPI table overflow handling

[Bjorn Helgaas <bjorn.helgaas@hp.com>]

On Sunday 11 January 2004 7:30 am, Jes Sorensen wrote:
> +++ linux-2.6.0-test11-ia64/arch/ia64/kernel/acpi.c	Sun Jan 11 05:15:22 2004
> -	if (acpi_table_parse_madt(ACPI_MADT_IOSAPIC, acpi_parse_iosapic) < 1)
> +	if (acpi_table_parse_madt(ACPI_MADT_IOSAPIC, acpi_parse_iosapic, 256) < 1)

The "256" looks like it's based on the "iosapic_lists[256]" definition.
We probably should introduce a #define for those cases (sorry, I should
have noticed this the first time).

> +++ linux-2.6.0-test11-ia64/arch/x86_64/kernel/acpi/boot.c	Sun Jan 11 05:31:58 2004
> +	result = acpi_table_parse_madt(ACPI_MADT_INT_SRC_OVR, acpi_parse_int_src_ovr, NR_IRQ_VECTORS);
>...
> +	result = acpi_table_parse_madt(ACPI_MADT_NMI_SRC, acpi_parse_nmi_src,
> +				       NO_IRQ_VECTORS);

Is NO_IRQ_VECTORS a typo for NR_IRQ_VECTORS?

> +++ linux-2.6.0-test11-ia64/drivers/acpi/numa.c	Thu Jan  8 02:49:24 2004
>  acpi_table_parse_srat (
>  	enum acpi_srat_entry_id	id,
> -	acpi_madt_entry_handler	handler)
> +	acpi_madt_entry_handler	handler,
> +	int max_entries)

Should "max_entries" be unsigned?  I notice you used unsigned types in the
implementation, i.e., "count".

Bjorn