[systemd-devel] [PATCH] arch: add crisv32

[Umut Tezduyar Lindskog]

Hi,

On Thu, Jul 3, 2014 at 1:09 PM, Lennart Poettering
<lennart at poettering.net> wrote:
> On Thu, 03.07.14 09:54, Umut Tezduyar Lindskog (umut.tezduyar at axis.com) wrote:
>
>> ---
>>  src/shared/architecture.c |    1 +
>>  1 file changed, 1 insertion(+)
>>
>> diff --git a/src/shared/architecture.c b/src/shared/architecture.c
>> index 9e0c3ef..7dd049a 100644
>> --- a/src/shared/architecture.c
>> +++ b/src/shared/architecture.c
>> @@ -116,6 +116,7 @@ Architecture uname_architecture(void) {
>>                  { "tilegx",     ARCHITECTURE_TILEGX   },
>>  #elif defined(__cris__)
>>                  { "cris",       ARCHITECTURE_CRIS     },
>> +                { "crisv32",    ARCHITECTURE_CRIS     },
>>  #else
>>  #error "Please register your architecture here!"
>>  #endif
>
> Applied, thanks!
>
> This makes me wonder though, was the "cris" identifier ever used by
> uname? Or was that simply incorrect cruft? If so, I'd prefer removing
> it?

Seems like this was a mistake. I checked up few products and uname
returns crisv32. I believe there is also crisv10 but I don't have a
product to verify it.

>
> One more thing though: in context of the file-hierarchy(7) man page I
> added to systemd I added support for querying various system paths via
> the new "systemd-path" tool. This tool wants to know the arch "tuple"
> used for the specific ABI, building on what Debian is using:
>
>    https://wiki.debian.org/Multiarch/Tuples
>
> I added support for detecting these ABIs to architecture.h, however, I
> couldn't fill in quite a number of them, becuase DEbian either didn't
> know these archs yet, and hence didn't list any tuple, or because I
> couldn't figure out how to detect the variety of ABIs used on the same
> architecture with just #ifdefs. For these architectures I currently
> added a "#error" line, which means the builds are broken on these archs,
> in the hope that people will quickly send me the appropriate patches.
>
> Currently the following archs are broken in this regard:
>
>           ppc64le
>           ppcle
>           hppa64
>           mips64
>           mips64le
>           32bit ARM (yay, at least 4 ABIs to choose from, well done!)
>           sh64
>           tilegx
>           cris
>
> Or with other words, systemd git will not build on your architecture,
> Umut!

Thanks for the heads up. I will send a patch when I stumble on it.

Umut

>
> I figure most of these archs don't have many fans who care, but I am
> pretty sure cris, tilegx and especially ARM is something people care
> about!
>
> Hence, please, send me a patch that tries to figure out a
> debian-compatible ABI tuple simply with C #ifdefs! (which I really hopen
> is even possible on those fucked up archs like ARM!) Otherwise 215 will
> not compile on your arch!
>
> Note that different Linux distributions tend to use different
> triplets. For example, on FEdora:
>
>    $ gcc -dumpmachine
>    x86_64-redhat-linux
>
> I am mostly interested in the Debian-version of the tuples, as they
> don't bake any company or distro name into the string.
>
> (Why this all? Well, in the long run I kinda hope we can bring some
> cross-distro sanity back to the library logic, and as the DEbian scheme
> of splitting per-arch library dirs up into tuple directories appears the
> most sensible, this is what we document in file-hierarchy(7).)
>
> Lennart
>
> --
> Lennart Poettering, Red Hat
> _______________________________________________
> systemd-devel mailing list
> systemd-devel at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/systemd-devel