Skip to Content.
Sympa Menu

svadev - Re: [svadev] Runtime calls and poolalloc failure

svadev AT lists.siebelschool.illinois.edu

Subject: Svadev mailing list

List archive

Re: [svadev] Runtime calls and poolalloc failure


Chronological Thread 
  • From: John Criswell <criswell AT illinois.edu>
  • To: Alex Miller <mille151 AT illinois.edu>
  • Cc: svadev AT cs.uiuc.edu
  • Subject: Re: [svadev] Runtime calls and poolalloc failure
  • Date: Mon, 26 Mar 2012 09:50:39 -0500
  • List-archive: <http://lists.cs.uiuc.edu/pipermail/svadev>
  • List-id: <svadev.cs.uiuc.edu>
  • Organization: University of Illinois

On 3/26/12 1:15 AM, Alex Miller wrote: Hey all,

In trying to get baggy bounds checking working, I've gotten stuck in a couple places:

1. In bringing up the baggy bounds checking runtime library, I've had to add a few stubs for functions like `boundscheckui_debug` to BBRuntime. However, I can't seem to find where calls to these functions are actually being inserted into the program. Where should I be looking to figure out why they're being added in and what they're supposed to be doing?

The _debug versions of the run-time checks are added by the DebugInstrumentation pass (lib/DebugInstrumentation/DebugInstrumentation.cpp).  This pass looks for SAFECode run-time checks and converts them into the _debug versions by adding extra debug information to the call.

For getting BBC up and running initially, I would recommend either disabling the DebugInstrumentation pass or implementing versions of the _debug functions that ignore the extra debug information information.


2. When trying to run poolalloc over a very small testcase[1], I keep on getting a "Pool allocating 0 global nodes!"[2] error that I can't seem to hunt down/understand.

I think that's just an informational message.  Does running poolalloc on the test case generate a poolallocated version of the program?

-- John T.




Archive powered by MHonArc 2.6.16.

Top of Page