Bug#1004495: 5.16.3-1~exp1 FTBFS on i386

[Salvatore Bonaccorso <carnil@debian.org>]

Control: tags -1 + confirmed
Control: forwarded -1 https://lore.kernel.org/lkml/20220114075756.838243-1-slyich@gmail.com/

Hi Daniel,

On Sat, Jan 29, 2022 at 10:23:21AM +0100, Daniel Baumann wrote:
> Package: linux
> Version: 5.16.3-1~exp1
> Severity: serious
> Tag: experimental
> 
> Hi,
> 
> unfortunately the last upload to experimental failed to build on i386.

Thank you Daniel from reporting, yes we are already aware of it, and for armel,
armhf (and will be for the other 32bit architectures). From yesterdays
conversation on IRC:

[23:49] < bwh> carnil: waldi: I suppose this is caused by https://salsa.debian.org/kernel-team/linux/commit/cf5238aa51bb6611bb7393ea6ec5f6f7c1a37711
[23:52] < bwh> and the preceding statement "idx = (rel->addend / sizeof(void *));" means the range of idx is twice as large on 32-bit, potentially requiring one more digit
[23:57] < bwh> also it seems like objtool can't actually handle cross-builds with differing word size...

The issue is there already since 5.16-rc1 but only got uncovered now with the
above change.

Regards,
Salvatore