XPost: linux.debian.ports.arm   
   From: arnd@arndb.de   
      
   On Thu, Oct 30, 2025, at 09:47, Peter Green wrote:   
   > On 30/10/2025 08:28, Arnd Bergmann wrote:   
   >> On Thu, Oct 30, 2025, at 06:39, Peter Green wrote:   
   >>> The problem is that arm64 kernels by default trap the armv6k memory   
   >>> barriers into the kernel and emulate them. Asside from being slow   
   >>> this can also cause hangs if the optimiser moves a barrier inside   
   >>> a load-exclusive/store-exclusive loop.   
   >> Right, this is very unfortunate, and I suspect that this is something   
   >> we should change in both the arm64 kernel and in the compiler. If   
   >> you can point me to specific userspace code that has this problem,   
   >> I can probably come up with a kernel patch to work around it,   
   >> at least if that userspace code isn't obviously wrong.   
   > https://github.com/rust-lang/rust/issues/53670   
      
   Thanks for the pointer! This strongly suggests that the bug is   
   in llvm itself, not in userspace applications or the kernel.   
   I think the armv7-a version of the reproducer shows the same bug,   
   though that happens to work. This is probably one of the cases   
   where we want a compiler fix first, but can then add a workaround   
   in the host kernel (enabling the instruction itself instead of   
   the emulation where possible, same as the Raspberry Pi downstream   
   kernel does).   
      
        Arnd   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)