Re: Bug#1127346: src:sprng: fails to migrate to testing for too long: FTBFS on armhf

Dirk Eddelbuettel <[email protected]>
Newsgroups gmane.linux.debian.ports.arm
Message-ID <[email protected]>
On 7 February 2026 at 23:19, Peter Green wrote:
| On 07/02/2026 08:45, Paul Gevers wrote:
| 
| > ∙ ∙ Missing build on armhf
| 
| The culrpit appears to be the following change.
| -DEB_HOST_ARCH := $(shell dpkg-architecture -qDEB_HOST_ARCH)
| -ifeq ($(DEB_HOST_ARCH),armhf)
| +arch            := $(shell dpkg-architecture -qDEB_BUILD_ARCH_CPU)
| +ifeq ($(arch),armhf)

Yes and no. You very selectively quoted here, the next two lines matter as we
only use this in a (desparate) attempt to get the compiler flags right. It is
not used anywhere else so it does not really matter.
 
| This change appears to be wrong in two ways.
| 
| 1. Checking the "host" arch (arch you are building for) not the "build" arch (arch you are building on) is generally what you want for compiler flags.
| 2. DEB_{BUILD|HOST}_ARCH_CPU on armhf is "arm" not "armhf".
| 
| With an arm porter hat on, I suggest reverting this change.

As you are an arm porter, could I ask you to instead log at one of the actual
failures eg via the last build attempt there at

https://buildd.debian.org/status/logs.php?pkg=sprng&ver=2.0a-17&arch=armhf

which gets us to

https://buildd.debian.org/status/fetch.php?pkg=sprng&arch=armhf&ver=2.0a-17&stamp=1767797253&raw=0

where we see

/tmp/ccTWsmvQ.s: Assembler messages:
/tmp/ccTWsmvQ.s:1735: Error: thumb conditional instruction should be in IT block -- `addcs r9,r9,#65536'
/tmp/ccTWsmvQ.s:1750: Error: thumb conditional instruction should be in IT block -- `addcs r8,r8,#65536'
/tmp/ccTWsmvQ.s:1785: Error: thumb conditional instruction should be in IT block -- `addcs r8,r8,#65536'
/tmp/ccTWsmvQ.s:1817: Error: thumb conditional instruction should be in IT block -- `addcs lr,lr,#65536'

I have no idea how to overcome this, so for now I turned armhf off.

This is an old package I added way back when we had no other parallel RNG for
R. It is no longer used by R (packages) (at either CRAN or in Debian) so I
could just remove it. Before I do so I would not mind sorting out why builds
that worked for so many years now fail.

Dirk

-- 
dirk.eddelbuettel.com | @eddelbuettel | [email protected]
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.