XPost: linux.debian.ports.arm   
   From: plugwash@debian.org   
      
   On 30/10/2025 08:28, Arnd Bergmann wrote:   
   > On Thu, Oct 30, 2025, at 06:39, Peter Green wrote:   
   >> On 29/10/2025 13:47, Arnd Bergmann wrote:   
   >>   
   >>   > 5. armv6k+vfpv3d16: Most ARMv6/v7 machines, including Raspberry   
   >>   >     Pi zero/1 but not OMAP2 (Nokia N800/N810). Loses THUMB2 support   
   >>   >     and v7/v8 CPU barriers among other minor differences.   
   >>   
   >> As maintainer of raspbian, I can say the memory barrier issue is   
   >> one of the bigger thorns in our side and is the reason we continue   
   >> to configure gcc for armv6 rather than armv6k.   
   >>   
   >> For the uninitiated, arm added memory barrier instructions in armv6k,   
   >> as instructions on the system coprocessor (aka CP15). When targetting   
   >> armv6k, compilers use these instructions to implement   
   >> acquire/release/seqcst atomics.   
   >>   
   >> With armv7-a, arm deprecated the CP15 barriers and introduced the   
   >> dmb instruction.   
   >>   
   >> 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   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)