Re: [PATCH v3 05/10] x86: remove HIGHMEM64G support

[H. Peter Anvin]

On February 27, 2025 8:51:59 AM PST, Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
>On Thu, 27 Feb 2025 at 07:41, H. Peter Anvin <hpa@xxxxxxxxx> wrote:
>>
>> One of the generations of kernel.org ran on a dual socket system with
>> 6 GiB RAM. It was a mess; basically it achieved less than 50% memory
>> utilization because of highmem.
>
>I'll be very very happy when HIGHMEM is gone completely, but yes,
>HIGHMEM64G was particularly painful.
>
>It was definitely used, and it worked better under some loads than
>others (large user footprints with lots of anonymous memory and little
>kernel side memory pressure), but it was not great in general.
>
>I suspect that absolutely everybody who ever used it switched to
>64-bit as soon as humanly possible and nobody has likely actively used
>it for a *long* time.
>
>Good riddance,
>
>            Linus

I wish that were true. 

At rPath I had to debug a customer machine where they insisted on a 32-bit OS because they were worried about the breakage possibilities of the compat layer. In retrospect we really didn't spend enough time/effort on making sure an all-32-bit userspace would run on a 64-bit kernel; little corner cases like autofs (my fault) that only popped up when all the system software was 32 bits...