I'm confident this is right, but I'm curious why you didn't do it the same way as in `longjmp`, by doing the original store in smaller blocks so that you don't clobber those registers in the first place. Set r1,r2,r3 to r8,r9,r10 and do a three-reg `stmia`; then set them to r11,sp,lr and do another one. One fewer instruction and less memory access. It's _more_ of a win here to do it like this than it was in `longjmp`, because you have one more low register free, so you can split the store into two 3-reg chunks instead of three 2-reg ones.

This approach will surely clobber the values that r4-r7 had on entry to `setjmp`? `setjmp` has to do two things with the values of callee-saved registers. It has to put them in the `jmp_buf` for `longjmp` to restore. But it _also_ has to preserve them itself, for when it returns directly to the caller.

This use of r1 ought to be accessing the value passed in to `longjmp` as a parameter. But that's not what's in r1 right now. The previous two `ldmia` both clobbered it. At the point of this instruction, the value in r1 is whatever value the jmp_buf had saved for r11.

ditto

Is it possible to have (link to) documentations / comments to help newbies like me on these asm?

Extra space before `[` here. This is already a problem for the x86-64 case, but it's not kosher to use plain identifiers here, or in any standard C header. They need to be `__`-prefixed in case a user has `#define opaque syntax-error` before `#include <setjmp.h>`. We should look into a clang-tidy check to enforce that.

The assembler accepts either, but I think we should follow the norm of using `#` for immediates in arm32 and aarch64 assembly.

Same issues here: `.w`,`!`, register range syntax, raw string, `#` prefix.

These warnings (pattern copied from other setjmp impl's in tree) are dumb. You'll get an obvious compiler diagnostic without it pretty quick! I'm going to remove them.