Re: [PATCH v3 5/6] gpu: nova-core: skip the IFR header if present

[John Hubbard <jhubbard@nvidia.com>]

On 4/15/26 6:32 PM, John Hubbard wrote:
> On 4/15/26 6:24 PM, Timur Tabi wrote:
>> On Wed, 2026-04-15 at 18:21 -0700, John Hubbard wrote:
>>> On 4/15/26 5:26 PM, Timur Tabi wrote:
>>> ...
>>>> +fn current_offset(dev: &device::Device, bar0: &Bar0) -> Result<usize> {
>>>> +    let signature = bar0.read(ga100::NV_PBUS_IFR_FMT_FIXED0).signature();
>>>> +
>>>> +    if signature == NV_PBUS_IFR_FMT_FIXED0_SIGNATURE_VALUE {
>>>> +        let fixed1 = bar0.read(ga100::NV_PBUS_IFR_FMT_FIXED1);
>>>> +
>>>> +        match fixed1.version() {
>>>> +            // Note: We do not actually expect to see v1 or v2 on these GPUs
>>>
>>> Then why even match for it? No need to "support" HW or FW that doesn't,
>>> and won't ever exist, right?
>>
>> As I tried to explain last time, I'm not 100% sure that we'll never see it.  It's just a
>> footnote, based on my examination of various documents on the IFR.  
> 
> I missed that, sorry. Let's either delete the comment, or else water
> it down to...well, I don't know what. Probably best to simply delete 
> it, otherwise someone will want to delete the code.
> 
> Either you think that there is *some* chance that this code is not
> perma-dead, or you don't. And it sounds like the latter.
> 

...ummm, let me reverse the polarity of that last sentence. :)

thanks,
-- 
John Hubbard