On Tuesday, March 2, 2021, Luke Kenneth Casson Leighton <l...@lkcl.net> wrote: > > > On Tuesday, March 2, 2021, Riccardo Mottola <riccardo.mott...@libero.it> wrote: > >> actually the original point is even for PPC32, note just PPC64. The >> configure check added by Adrian in Firefox checks if the compiler >> accepts -maltivec and just enables it in the build. >> However, this assumption is not correct and causes issues as explained >> in my previous mail. > > ouch. seems like an autoconf test is needed, at least. and an upstream bugreport to gcc. > > just to confirm: that's definitely "setting machine to capabilities that the machine doesn't have, then requesting that feature and gcc 10 says 'ok'" yes? > > i do not know the exact machine, let us assume it is -mg3. > > the options being passed are "gcc -mg3 -maltivec" and this should definitely cause gcc to raise an error, is that correct?
or, is it: * just -mnoaltivec * no specific setting of machine type * VMX instructions still get introduced whilst i do not know if gcc rejects inline VSX assembly if -mnoaltivec is given, i have a sneaking suspicion that this could be something not to do with gcc itself but with e.g. recent libc6 proliferation of inline assembly variants of functions such as strncpy. are you able to send a gdb stacktrace here to the list and also a disasm dump at the PC showing which instruction is being attempted? this will tell us what function is going awry and we can then ping the right people. l. -- --- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
