Re: [PATCH net-next 0/5] Remove uses of kmap_atomic()

[Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>]

On 11/22/2022 11:34 PM, Leon Romanovsky wrote:
> On Tue, Nov 22, 2022 at 01:06:09PM -0800, Anirudh Venkataramanan wrote:
>> On 11/22/2022 10:50 AM, Jakub Kicinski wrote:
>>> On Tue, 22 Nov 2022 13:29:03 +0200 Leon Romanovsky wrote:
>>>>>    drivers/net/ethernet/sun/cassini.c            | 40 ++++++-------------
>>>>>    drivers/net/ethernet/sun/sunvnet_common.c     |  4 +-
>>>>
>>>> Dave, Jakub, Paolo
>>>> I wonder if these drivers can be simply deleted.
>>>
>>> My thought as well. It's just a matter of digging thru the history,
>>> platform code and the web to find potential users and contacting them.
>>
>> I did a little bit of digging on these two files. Here's what I found.
>>
>> For the cassini driver, I don't see any recent patches that fix an end user
>> visible issue. There are clean ups, updates to use newer kernel APIs, and
>> some build/memory leak fixes. I checked as far back as 2011. There are web
>> references to some issues in kernel v2.6. I didn't see anything more recent.
>>
>> The code in sunvnet_common.c seems to be common code that's used by
>>
>> [1] "Sun4v LDOM Virtual Switch Driver" (ldmvsw.c, kconfig flag
>> CONFIG_LDMVSW)
>>
>> [2] "Sun LDOM virtual network driver" (sunvnet.c, kconfig flag
>> CONFIG_SUNVNET).
>>
>> These two seem to have had some feature updates around 2017, but otherwise
>> the situation is the same as cassini.
> 
> If there is a pole to delete them, I vote for deletion. :)

Perhaps I should put out a patchset that does this. That would give 
users or anyone who cares another opportunity to chime in. If we hear 
nothing, then we're good.

I am thinking I'll still include the conversions for cassini.c and 
sunvnet_common.c in my v2 series. In case we don't end up deleting these 
drivers for some reason, they'd at least not be using kmap_atomic() anymore.

Ani