What Do You Think? JDK-8213084: Rework and enhance Print[Opto]Assembly output
Andrew Dinn
adinn at redhat.com
Fri Nov 30 10:30:44 UTC 2018
I am entirely in agreement with Andrew Haley:
The blank lines in the original format are critical to conspicuity of
independent groups of information. Gestalt matters.
The additional detail (keywords, brackets offset, insn bytes) add almost
zero /useful/ information, effectively blotting the original signal with
noise.
Also, I'm also dubious over the utility of providing a hex byte dump of
the code when no hsdis.so is available. I say that having written a
simulator for AArch64 (including, therefore, a decode unit) and coded &
debugged a lot of AArch4 and x86 assembler and, occasionally, machine
code. So, I'm probably one of your most promising potential customers
for such a feature and I'm (at best) lukewarm.
I'll allow that there /may/ be situations where one could arrange for
PrintAsssembly to be enabled on a java command line but could not
arrange for an hsdis.so to be installed in the LD_LIBRARY_PATH yet I'm
struggling to think of one. Faced with the task of decoding the hex I'd
much rather spend time working on whatever problems were involved in
building and/or installing the library.
regards,
Andrew Dinn
-----------
Senior Principal Software Engineer
Red Hat UK Ltd
Registered in England and Wales under Company Registration No. 03798903
Directors: Michael Cunningham, Michael ("Mike") O'Neill, Eric Shander
More information about the hotspot-compiler-dev
mailing list