Re: dmidecode
On 3 Aug 2021, at 23:42, Paul Wise <pabs@debian.org> wrote:
>
> On Tue, 03 Aug 2021 20:14:57 +0200 Jörg Frings-Fürst wrote:
>
>> because of the bugreport[1] I would like to ask here if it is useful to
>> compile dmidecode also for riscv64?
>
> Apparently the SMBIOS spec v3.3.0 and later includes RISC-V support:
>
> https://github.com/riscv/riscv-smbios/blob/master/RISCV-SMBIOS.md
>
> I've no idea if dmidecode meaningfully works on RISC-V hardware though?
I’m unaware of any systems that provide it. The whole EDK2+ACPI+SMBIOS
ecosystem remains rather far off for RISC-V at this point, especially
given ACPI will only include bindings for the future generation of
interrupt controllers and thus will never be supported for the current
generation of hardware. If it compiles then it might as well be
enabled, but this hardly seems essential, nor worth Ubuntu carrying a
patch for.
Looking at dmidecode itself, the primary reason it shouldn’t be
Architecture: any is it assumes unaligned access and little-endianness
by default unless you put -DALIGNMENT_WORKAROUND/-DBIGENDIAN in CFLAGS
manually, which is atrocious software engineering practice in many
respects... and for added fun it defines its own u64 type as a struct
containing a pair of u32’s (also manually typedef’ed to unsigned int)
whose order depends on BIGENDIAN.
Jess
Reply to: