proccontrol: fix process attachment without an exe

[cuviper]

This fixes two cases where the low-level process was assumed to have a valid executable path, but it may not if a process was attached without ever specifying the exe.

• In computeAddrWidth we can rewrite the test to work with both big- and little-endian auxv.
• In Codegen, we can get the executable path from the libraries() pool instead.

Fixes #146

[cuviper]

cc @mcfadden8 -- please double-check that this still accomplishes what you needed for PPC64LE.

[mcfadden8]

These changes appear to be right for PPC64LE. I'll download the branch and give the fix a try on the PPC64LE we have here.

[mcfadden8]

I am having difficulty compiling the latest version of the testsuite on PPC64LE. I am seeing the following compilation error:

$ make
[  0%] Built target testB_static
[  0%] Built target testB
[  0%] Built target testA_static
[  0%] Building CXX object CMakeFiles/testSuite.dir/src/JUnitOutputDriver.cpp.o
/g/g0/martymcf/devel/src/testsuite_9.2/src/JUnitOutputDriver.cpp: In constructor 'JUnitOutputDriver::JUnitOutputDriver(void*)':
/g/g0/martymcf/devel/src/testsuite_9.2/src/JUnitOutputDriver.cpp:13:50: error: 'getpid' was not declared in this scope
     results_log_name << "test_results" << getpid() << ".xml";
                                                  ^
make[2]: *** [CMakeFiles/testSuite.dir/src/JUnitOutputDriver.cpp.o] Error 1
make[1]: *** [CMakeFiles/testSuite.dir/all] Error 2
make: *** [all] Error 2

Has anyone else seen this?

I'll write up a separate PR in testsuite if I determine that this isn't somthing strange in my environment.

[cuviper]

Yeah, I pulled the testsuite and I see the same getpid() error. That file needs to include the right headers to get that declaration. I'll just opened dyninst/testsuite#12.

[mcfadden8]

Oops. I was writing dyninst/testsuite#11 while you were opening yours

[mcfadden8]

I confirmed that I am able to successfully run the proccontrol testsuite on ppc64le against a version of proccontrol with the above patches applied. Thanks for catching/fixing this.

uname -a
Linux rzmist1 3.16.0-30-generic #40~14.04.1-Ubuntu SMP Thu Jan 15 17:42:36 UTC 2015 ppc64le ppc64le ppc64le GNU/Linux


./runTests -proccontrol
TEST                       COMP   OPT  ABI MODE     THREAD  LINK    PIC     RESULT
pc_addlibrary              g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_breakpoint              g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_detach                  g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_fork                    g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_fork_exec               g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_groups                  g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_irpc                    g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_launch                  g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_library                 g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_singlestep              g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_stat                    g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_temp_detach             g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_terminate               g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_terminate_stopped       g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_thread                  g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_thread_cont             g++    none 64  create   SPST    dynamic nonPIC  PASSED
pc_tls                     g++    none 64  create   SPST    dynamic nonPIC  PASSED

[cuviper]

That's ok, I just cross-referenced it. 😄

[cuviper]

Thanks for confirming! Let's merge it then, and I'll cherry-pick it for master too.