On Fri, Aug 28, 2020 at 04:10:10PM +0200, John Paul Adrian Glaubitz wrote: > Not sure whether it's an actual problem because the majority of users with > a 32-bit kernel will be on Mac hardware, so the number of fixed systems > will certainly outweigh the number of systems which can't be debugged.
That could be. I have no idea what percentage of powerpc 32 bit systems are mac versus other systems. The ones I have worked with are likely not a representative sample at all given no macs were involved. I am not saying don't make the change, I am just asking if the potential impact is worth it. It could very well be the answer is yes. > I assume the bug should be fixed by the person who introduced the bug. I wouldn't. The code may very well predate the VMAP_STACK support. Adding VMAP_STACK would be unrelated to the broken code, but does affect it of course in an unfortunate way. So you probably need the person that wrote the now broken assembly code to fix it, even though when it was written there was probably nothing wrong with it. -- Len Sorensen
