RFR: 8225429: Regression in disassembly quality

[Andrew Haley]

On 6/6/19 9:22 PM, John Rose wrote:

> Effective diagnosis and performance tuning requires accurate,
> informative disassembly.  I would support [enabling] the assembler
> comment capture, whenever disassembly is enabled.  If there's a risk
> of disturbing program dynamics with the extra footprint, perhaps it
> could be broken out as a separate diagnostic option.
> 
> More disassembly information is better, even in product mode!

OK, I'll look at that. There are multiple issues with assembler
comments, so I'll try to collect them up. If we're not to slow down
code generation we'll need a fast test-and-branch so that we can gate
assembler comment capture. That shouldn't be too hard.

-- 
Andrew Haley
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
https://keybase.io/andrewhaley
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671