427135 - Add super-H(sh3,4) architecture support

Status: RESOLVED FIXED

(NSS :: Build, enhancement, P2)

Description

[Kai Engert (:KaiE:)]

Attachment: nss-3.11-sh-linux.patch

This patch adds support super-H(sh3,sh4) architecture.
This is forward ported patch for version nss-3.11.99.4-1
getting from http://www.sh-linux.org/ .
 
The original patch is for nss-3.11, but it can
also apply to nss-3.11.99.4-1 and compile.

Originally reported in https://bugzilla.redhat.com/show_bug.cgi?id=437676

Comment 1

[Nelson Bolyard (seldom reads bugmail)]

OK, I'll bite.  What are sh3 and sh4?

Comment 2

[Kai Engert (:KaiE:)]

I suspect it's this: http://en.wikipedia.org/wiki/SuperH

Comment 3

[Nelson Bolyard (seldom reads bugmail)]

Is this patch all that is needed?  
Or is it the first of many to come?

Who will primarily support issues with NSS on Super-H? 
Raúl ?
Kai  ?
Reed ?

Comment 4

[Nelson Bolyard (seldom reads bugmail)]

To expand on the questions in the preceding comment, the NSS team's recent 
experience with WinCE has shown us that we CANNOT claim to support any new 
platforms without an NSS tinderbox for that platform.  We are at our 
capacity to run our own tinderboxes, which means that tinderboxes for new 
platforms must be run by the champions of those new platforms. 

So, the question I should have asked in comment 3 is: 

Who will supply the NSS tinderbox system, and keep it running, for Super-H?

It is also HIGHLY desirable that for each new platform, we have one or more
developers who will commit to support that platform.

Comment 5

[Nelson Bolyard (seldom reads bugmail)]

Kai, Reed,
On whose behalf did you file this bug and request patch review?
Are either of you personally involved in Super-H work?
Is there someone who can answer my review questions on behalf of the 
Super-H developers?

Comment 6

[Reed Loden [:reed]]

Raúl Porcel may know somebody.

Comment 7

[Raúl Porcel]

(In reply to comment #3)
> Is this patch all that is needed?  
> Or is it the first of many to come?
> 
This patch should be enough.

> Who will primarily support issues with NSS on Super-H? 
> Raúl ?
> Kai  ?
> Reed ?
This patch just adds support for SuperH(i.e, the arch is recognized), i don't think it will create issues. Kai is just a Fedora/Redhat developer with nothing to do with SuperH(Redhat/Fedora doesn't support SH). I'm a Gentoo developer in the Gentoo SH team(Gentoo supports SH). However I don't think i would be able to support issues with this, since my knowledge is very limited. Still, the patch is very minimal, I don't think it hurts, but of course thats your decision.

(In reply to comment #4)
> To expand on the questions in the preceding comment, the NSS team's recent 
> experience with WinCE has shown us that we CANNOT claim to support any new 
> platforms without an NSS tinderbox for that platform.  We are at our 
> capacity to run our own tinderboxes, which means that tinderboxes for new 
> platforms must be run by the champions of those new platforms. 
> 
> So, the question I should have asked in comment 3 is: 
> 
> Who will supply the NSS tinderbox system, and keep it running, for Super-H?
I'm not able to provide that, sorry.

> It is also HIGHLY desirable that for each new platform, we have one or more
> developers who will commit to support that platform.
As I said avobe, my knowledge on C and programming is very limited, so I don't think I could help too much.

(In reply to comment #5)
> Kai, Reed,
> On whose behalf did you file this bug and request patch review?
Kai filed the bug on behalf of CHIKAMA Masaki, which is a Fedora SH user(check the link on comment #1), he just forwarded the bug here, as explained on that comment.
As for Reed, he requested review on my behalf, since I saw this bug that needed a review.

> Are either of you personally involved in Super-H work?
As I said avobe, I'm a Gentoo developer and I'm in the Gentoo SH team.
> Is there someone who can answer my review questions on behalf of the 
> Super-H developers?
Guess I already answered those :)

Comment 8

[Nelson Bolyard (seldom reads bugmail)]

Raul, with this patch applied to your own copy of the sources, are you able
to build NSS in its entirety for SuperH?  

Does superH run full blown gentoo linux?  Or is gentoo linux merely a 
cross-development environment for whatever OS is used by SuperH?

Comment 9

[Nelson Bolyard (seldom reads bugmail)]

The question in comment 8 is really the crucial question.

If SuperH is just another Linux platform, running the same full-blown Linux
as runs on countless X86 laptops, X86 servers, S390 servers, any many other
CPUs, then I have no worries about it.  In that case we can apply this patch
right away.

But if SuperH is more like WinCE (Windows Mobile), being a lot like desktop
windows, but lacking in many MANY features upon which NSS and NSPR rely, 
then adding support for it is a big deal.  

So, please let us know which of these situations is most like that of SuperH.

Comment 10

[Raúl Porcel]

Sorry for taking a bit to answer, but i was trying this, since the patch doesn't seem to apply anymore.

(In reply to comment #8)
> Raul, with this patch applied to your own copy of the sources, are you able
> to build NSS in its entirety for SuperH?  
Not with this one, I made another i will attach.

> 
> Does superH run full blown gentoo linux?  Or is gentoo linux merely a 
> cross-development environment for whatever OS is used by SuperH?
Yes, SuperH is able to run a full linux system, at least the sh3 and sh4{,a} processors.
Have a look:
sh4-unknown-linux-gnu-gcc -o Linux2.6_sh4_sh4-unknown-linux-gnu-gcc_glibc_PTH_OPT.OBJ/mangle.o -c -O2 -fPIC -DLINUX1_2 -D_XOPEN_SOURCE -DLINUX2_1  -ansi -Wall -Werror-implicit-function-declaration -Wno-switch -pipe -DLINUX -Dlinux -D_POSIX_SOURCE -D_BSD_SOURCE -DHAVE_STRERROR -DXP_UNIX -DSHLIB_SUFFIX=\"so\" -DSHLIB_PREFIX=\"lib\" -UDEBUG -DNDEBUG -D_REENTRANT -DNSS_ENABLE_ECC -DUSE_UTIL_DIRECTLY -I../../../../dist/Linux2.6_sh4_sh4-unknown-linux-gnu-gcc_glibc_PTH_OPT.OBJ/include -I../../../../dist/public/nss -I../../../../dist/private/nss -I/usr/include/nspr -I../../../../dist/Linux2.6_sh4_sh4-unknown-linux-gnu-gcc_glibc_PTH_OPT.OBJ/include/dbm -O2 -pipe mangle.c


(In reply to comment #9)
> The question in comment 8 is really the crucial question.
> 
> If SuperH is just another Linux platform, running the same full-blown Linux
> as runs on countless X86 laptops, X86 servers, S390 servers, any many other
> CPUs, then I have no worries about it.  In that case we can apply this patch
> right away.
Yes, it is :)

> 
> But if SuperH is more like WinCE (Windows Mobile), being a lot like desktop
> windows, but lacking in many MANY features upon which NSS and NSPR rely, 
> then adding support for it is a big deal.
WinCE can run on SuperH as well, but this patch is for Linux :)
> 
> So, please let us know which of these situations is most like that of SuperH.

Comment 11

[Raúl Porcel]

Attachment: nss.patch

However, since NSS depends on NSPR, you may want until bug 427136 is fixed to checkin this.

Comment 12

[Nelson Bolyard (seldom reads bugmail)]

Attachment: Alternative patch v1, simplify Linux.mk (checked in)

Raúl, please confirm that this patch works for you.
Christophe, please review.

Comment 13

[Christophe Ravel]

Comment on attachment 375344 [details] [diff] [review]
Alternative patch v1, simplify Linux.mk (checked in)

r+=christophe

Comment 14

[Raúl Porcel]

(In reply to comment #12)
> Created an attachment (id=375344) [details]
> Alternative patch v1, simplify Linux.mk (untested)
> 
> Raúl, please confirm that this patch works for you.
> Christophe, please review.
It works, thanks.

Comment 15

[Nelson Bolyard (seldom reads bugmail)]

Comment on attachment 375344 [details] [diff] [review]
Alternative patch v1, simplify Linux.mk (checked in)

Checking in Linux.mk; new revision: 1.36; previous revision: 1.35