portindex still broken in Rosetta: tclsh8.6(642,0xb0221000) malloc: *** error for object 0x1c00040: pointer being freed was not allocated

[barracuda156]

This is not on 10.6 ppc, this is on 10.6.8 on a physical x86_64 machine, MacPorts 2.10.4:

Creating port index in /opt/local/var/macports/sources/macos-powerpc.org/MacPorts/ports
tclsh8.6(642,0xb0221000) malloc: *** error for object 0x1c00040: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb0221000) malloc: *** error for object 0x1d005d0: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb03a7000) malloc: *** error for object 0x80a05910: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb0429000) malloc: *** error for object 0x1c00040: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb03a7000) malloc: *** error for object 0x6072c0: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb0325000) malloc: *** error for object 0x80a05f70: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb02a3000) malloc: *** error for object 0x1c00170: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb03a7000) malloc: *** error for object 0x1c00170: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb02a3000) malloc: *** error for object 0x6072c0: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb04ab000) malloc: *** error for object 0x80a05f70: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb03a7000) malloc: *** error for object 0x500ed0: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb03a7000) malloc: *** error for object 0x80a05f30: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb0429000) malloc: *** error for object 0x110c10: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb0221000) malloc: *** error for object 0x110bd0: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
tclsh8.6(642,0xb0325000) malloc: *** error for object 0x1b00210: pointer being freed was not allocated
*** set a breakpoint in malloc_error_break to debug
Command failed: /opt/local/bin/portindex /opt/local/var/macports/sources/macos-powerpc.org/MacPorts/ports
Killed by signal: 10

I do not have this problem on any native powerpc, at least so far (including both 10a190 and hacked 10.6.8 running natively). It also does not seem to happen on 10a190 on i386 when MacPorts is built for i386.

Now, one may ask why at all build MacPorts itself as ppc binaries (or course one can still build ports for ppc while MacPorts is built as x86_64); there is one convenience reason for that: platform darwin powerpc and ${os.arch} eq "powerpc" work correctly (in a sense of how they are usually used) only when MacPorts is built as ppc, and changing almost every instance of that to build_arch-based conditions is a lot of patching. However, this is not a big deal as such, but the issue looks like a bug in the code introduced in 2.8.x or so.

System info:

sergeys-mini:~ svacchanda$ uname -a
Darwin sergeys-mini.lan 10.8.0 Darwin Kernel Version 10.8.0: Thu Sep  6 11:44:53 AST 2012; root:xnu-1504.15.3/BUILD/obj//RELEASE_X86_64 x86_64
sergeys-mini:~ svacchanda$ sw_vers
ProductName:	Mac OS X
ProductVersion:	10.6.8
BuildVersion:	10K549
sergeys-mini:~ svacchanda$ port -v 
MacPorts 2.10.4

[jmroot (Joshua Root)]

#69301