On Mon, Nov 28, 2022 at 1:07 PM matoro <matoro_mailinglist_kernel@xxxxxxxxx> wrote: > > That seemed to do the trick! Looks like it needs support added to > golang.org/x/sys/unix: https://dpaste.com/4K99E4DWH.txt > > This is something I can look into on my own though. I might come back > to ask about the asm parts, but will try and figure it out. > > So does this mean it is a GC bug? It's a GC bug but it's probably going to require a parisc specific fix, likely in runtime/stack.c though possibly elsewhere as well. Ian > -------- Original Message -------- > Subject: Re: Correct goarch.sh settings for gccgo on hppa? > Date: 2022-11-28 15:20 > From: Ian Lance Taylor <iant@xxxxxxxxxx> > To: Helge Deller <deller@xxxxxx> > > On Sun, Nov 27, 2022 at 12:19 AM Helge Deller <deller@xxxxxx> wrote: > > > > I don't know if it's relevant, but on hppa the stack grows upwards, > > while on all(?) other platforms the stack grows downwards. > > Maybe something is missing, esp. since setcontext() is involved where > > a new context including stack is set and this leads to the "corrupt > > stack" ? > > Interesting point and quite possibly relevant. See the code in > runtime/stack.c. More generally, see if the tests behave better with > the environment variable GOGC=off. That disables the garbage > collector. > > Ian
