[Lldb-commits] [lldb] [lldb] New expedited register specfication for unavailable regs (PR #193894)

[Jason Molenda via lldb-commits]

================
@@ -274,23 +295,31 @@ bool GDBRemoteRegisterContext::ReadRegisterBytes(const RegisterInfo *reg_info) {
           success = false;
         else {
           // Read the containing register if it hasn't already been read
-          if (!GetRegisterIsValid(prim_reg))
+          if (GetRegisterIsUnfetched(prim_reg))
             success = GetPrimordialRegister(prim_reg_info, gdb_comm);
----------------
jasonmolenda wrote:

I think the real question we need to decide:  If the remote stub says a register cannot be read, should we allow the user to retry that by doing another register read?   If the remote stub provides partial bytes, should we allow the user to retry that?

The only argument for partial-bytes that I can imagine is a g/G packet that is shorter than the entire register context size.  But we're imagining a g/G that has PART of one register.  When hopefully what we would see is e.g. a g/G that gives you all the GPRs but doesn't include the FPRs, and then when the user tries to read an FPR, we fetch it with p/P.  It seems very bad if we get partial bytes for a register.

https://github.com/llvm/llvm-project/pull/193894