FIX #1418 (partially)
authorSimon Marlow <simonmar@microsoft.com>
Wed, 13 Jun 2007 10:29:28 +0000 (10:29 +0000)
committerSimon Marlow <simonmar@microsoft.com>
Wed, 13 Jun 2007 10:29:28 +0000 (10:29 +0000)
commit23e5985c3db852981d527d10d6a6271688049790
tree7b87dd13cdbbebb56be5e530fe524bd5694a2d22
parent62d948405f6b9a95fe4b31b7cffa387e5425d6db
FIX #1418 (partially)
When the con_desc field of an info table was made into a relative
reference, this had the side effect of making the profiling fields
(closure_desc and closure_type) also relative, but only when compiling
via C, and the heap profiler was still treating them as absolute,
leading to crashes when profiling with -hd or -hy.

This patch fixes up the story to be consistent: these fields really
should be relative (otherwise we couldn't make shared versions of the
profiling libraries), so I've made them relative and fixed up the RTS
to know about this.
compiler/cmm/CmmParse.y
compiler/codeGen/CgInfoTbls.hs
includes/InfoTables.h
rts/Printer.c
rts/ProfHeap.c
rts/RetainerProfile.c
rts/RetainerSet.c