Hello Simon, Thank you for the insight.I have dug into this a bit more, and I believe the test failure is not due to IBM vs IEEE floating point differences. A few key observations: 1. The failure also occurs on arm64, which has always used IEEE . This strongly suggests that the issue is not specific to IBM format. Also most probably the recent transitions are for long double and here btas is using double. 2. I have also checked from the sbuild shell and it seems like that it is using glibc 2.41 with the IEEE long double ABI on ppc64le and toolchain macros also indicate IEEE compliance.
Given this, I doubt if adding a conditional check for IBM format will work here.
On 19/09/25 10:04, Simon Richter wrote:
Is that IEEE vs IBM floating point, or x87 vs IEEE?In the former case, I wonder if we can make this conditional on using IBM floating point format, given that Debian is also moving the ppc64le port to IEEE.Simon