Sergey Bugaev, le ven. 26 mai 2023 12:49:55 +0300, a ecrit: > On Fri, May 26, 2023 at 4:32 AM Samuel Thibault <samuel.thiba...@gnu.org> > wrote: > > I'm trying to start a 64bit Debian/Hurd but this fails very early. > > > > 0xffffffff8107a712 is the *read_exec call in exec_load, which does > > happen fine, until read_exec calls copyout, which thus apparently fails > > somehow. > > > > Does that ring a bell? > > So gnumach faults while loading the ELFs, before ever jumping to userspace?
Yes. > And the addresses seem well above the 4 GB limit; Which addresses? > so this is probably with Luca's latest patchset. I have not tried > that, so no, it does not ring a bell, sorry. Maybe try what I've been > doing: without that patchset but with the BRK_START hack. I'm still using the BRK_START hack. Samuel
