[PATCH] D91878: [CSSPGO] Pseudo probe encoding and emission.

Tue Dec 8 21:53:33 PST 2020

hoy added inline comments.

================
Comment at: llvm/lib/MC/MCAsmStreamer.cpp:2135-2136
+     << Attr;
+  // Emit inline stack like
+  //  @ GUIDDirectCaller:11 @ GUIDCaller:1 @ GUIDmain:3
+  if (!InlineStack.empty()) {
----------------
hoy wrote:
> wmi wrote:
> > It is a reverse of sequence as the context string in the profile. Is it intentional? 
> Yes, it's intentional. We feel that it's easier to read when the inline context gets long. The decoding tool (https://reviews.llvm.org/D92334) also prints in the reverse order. What do you think? Would you perfer the profile context order? 
I might be wrong. The decoding tool prints in the profile context (caller-callee) order. We should do the same here.

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D91878/new/

https://reviews.llvm.org/D91878