[compiler-rt][hwasan] Add newline between record_addr lines on frame record dumps

[leonardchan]

Summary

If SymbolizePC failes, it's possible for the newline to not be emitted.

[pcc]

This isn't how the output looks on Android. Are you sure this isn't a Fuchsia-specific bug in the output formatting?

[vitalybuka]

In D103845#2804441, @pcc wrote:

This isn't how the output looks on Android. Are you sure this isn't a Fuchsia-specific bug in the output formatting?

I don't see from the code how android can insert \n if SymbolizePC failed

[leonardchan]

In D103845#2804441, @pcc wrote:

This isn't how the output looks on Android. Are you sure this isn't a Fuchsia-specific bug in the output formatting?

I think the newline gets added to the frame description only if Symbolizer::GetOrInit()->SymbolizePC(pc) is nonnull, so if it fails then no newline is added.

[pcc]

In D103845#2806211, @leonardchan wrote:

In D103845#2804441, @pcc wrote:

This isn't how the output looks on Android. Are you sure this isn't a Fuchsia-specific bug in the output formatting?

I think the newline gets added to the frame description only if Symbolizer::GetOrInit()->SymbolizePC(pc) is nonnull, so if it fails then no newline is added.

I see. I guess we never tested the code path where SymbolizePC returned null on Android.

I think I would prefer doing this in a slightly different way where the \n gets added in the Printf call on line 244. Then you can remove the \n on line 239 and you shouldn't need the else clause.

Where does the {{{bt:0:0x214178819a54}}}come from on Fuchsia? Is it printed to the console as a side effect of calling SymbolizePC?

[eugenis]

In D103845#2806211, @leonardchan wrote:

I think the newline gets added to the frame description only if Symbolizer::GetOrInit()->SymbolizePC(pc) is nonnull, so if it fails then no newline is added.

This makes sense to me. This boils down to ListOfModules being unimplemented on Fuschia, right?

[leonardchan]

Oh, actually @pcc was right on closer inspection. This is a misunderstanding on my end. So the {{bt:...}}} comes from

constexpr const char *kFormatFrame = "{{{bt:%u:%p}}}";

defined in sanitizer_symbolizer_fuchsia.h and this is only used in the definition for RenderFrame in sanitizer_symbolizer_markup.cpp:

void RenderFrame(InternalScopedString *buffer, const char *format, int frame_no,
                 uptr address, const AddressInfo *info, bool vs_style,
                 const char *strip_path_prefix, const char *strip_func_prefix) {
  CHECK(!RenderNeedsSymbolization(format));
  buffer->append(kFormatFrame, frame_no, address);
}

and it's here where a newline isn't being added. The format isn't used at all. So it might be more appropriate to either just add a newline to those formats (or this append) or go ahead and add the newline to the Printf call on line 244.

[pcc]

Thanks for tracking this down. It seems best to change the Printf call to add the newline, since otherwise it looks like you'd be adding a spurious newline to the other callers of RenderFrame on Fuchsia.

[pcc]

LGTM

compiler-rt/lib/hwasan/hwasan_report.cpp

for (uptr i = 0; i < frames; i++) {
     uptr pc = record & pc_mask;
     frame_desc.append("  record_addr:0x%zx record:0x%zx",
                       reinterpret_cast<uptr>(record_addr), record);
     if (SymbolizedStack *frame = Symbolizer::GetOrInit()->SymbolizePC(pc)) {
       RenderFrame(&frame_desc, " %F %L\n", 0, frame->info.address, &frame->info,
                   common_flags()->symbolize_vs_style,
                   common_flags()->strip_path_prefix);
       frame->ClearAll();
+    } else {
+      frame_desc.append("\n");
     }
     Printf("%s", frame_desc.data());
     frame_desc.clear();
   }
 }
 
 // Returns true if tag == *tag_ptr, reading tags from short granules if
 // necessary. This may return a false positive if tags 1-15 are used as a