I don't think that it is the deoptimization jump table as it happens for non-optimized code, but you are probably right that we try to disassemble some data by accident. The strange thing is that it only happens on one of the buildbots while running tests. It cannot be reproduced neither locally nor by logging into the actual buildbot machine and running that single test.
Regards, Søren On Wed, Jun 29, 2011 at 16:03, Alexandre Rames <[email protected]>wrote: > Hello, > > I recently spotted that the disassembler can fail when trying to > disassemble the deoptimization jump table. Would that be the cause of a bug > you are looking at? > This is because some data is interleaved within the code (via > Assembler::dd) in LCodeGen::GenerateDeoptJumpTable. > Currently there is no mechanism to tell the disassembler that there is some > data inlined, so it tries to disassemble it and fails if the debug mode is > enabled. > > Alexandre > > On Wed, Jun 29, 2011 at 2:38 PM, <[email protected]> wrote: > >> LGTM >> >> >> http://codereview.chromium.**org/7250002/<http://codereview.chromium.org/7250002/> >> >> -- >> v8-dev mailing list >> [email protected] >> http://groups.google.com/**group/v8-dev<http://groups.google.com/group/v8-dev> >> > > -- v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev
