Re: [patch 1/2] Incorrect PCI interrupt assignment on ES7000 for platform GSI

[Tonnerre]

Salut,

On Mon, Sep 13, 2004 at 07:50:59PM -0600, Natalie.Protasevich@xxxxxxxxxx wrote:
> diff -puN arch/i386/kernel/acpi/boot.c~mypatch arch/i386/kernel/acpi/boot.c
> --- linux/arch/i386/kernel/acpi/boot.c~mypatch	2004-09-13 14:08:21.192015024 -0600
> +++ linux-root/arch/i386/kernel/acpi/boot.c	2004-09-13 14:10:51.457171248 -0600
> @@ -442,6 +442,7 @@ int acpi_gsi_to_irq(u32 gsi, unsigned in
>  unsigned int acpi_register_gsi(u32 gsi, int edge_level, int active_high_low)
>  {
>  	unsigned int irq;
> +	unsigned int plat_gsi;
>  
>  #ifdef CONFIG_PCI
>  	/*
> @@ -463,10 +464,10 @@ unsigned int acpi_register_gsi(u32 gsi, 
>  
>  #ifdef CONFIG_X86_IO_APIC
>  	if (acpi_irq_model == ACPI_IRQ_MODEL_IOAPIC) {
> -		mp_register_gsi(gsi, edge_level, active_high_low);
> +		plat_gsi = mp_register_gsi(gsi, edge_level, active_high_low);
>  	}
>  #endif
> -	acpi_gsi_to_irq(gsi, &irq);
> +	acpi_gsi_to_irq(plat_gsi, &irq);
>  	return irq;
>  }
>  EXPORT_SYMBOL(acpi_register_gsi);

Looking at that,  won't that cause problems if  we don't have IOAPIC?
Then you end up using an undefined value as GSI.

			    Tonnerre

Attachment: signature.asc
Description: Digital signature