Re: [Tinycc-devel] Preprocessor performance tuning

[Sergey Korshunoff]

>> And "valgrind  --leak-check=full <cmd>" provides no help?
> I just discovered. A help fron the tcc (MEM_DEBUG) was expected.

A quick result (on the start of compilation):
CC=valgrind --leak-check=full i386-tcc -D__GNUC__=2
-D__GNUC_MINOR__=95 -fold-struct-init-code -bench
step 1: compile & link
==11033== Memcheck, a memory error detector
==11033== Copyright (C) 2002-2013, and GNU GPL'd, by Julian Seward et al.
==11033== Using Valgrind-3.9.0 and LibVEX; rerun with -h for copyright info
==11033== Command: i386-tcc -D__GNUC__=2 -D__GNUC_MINOR__=95
-fold-struct-init-code -bench -o vmlinux -nostdlib -static -Wl
==11033== Source and destination overlap in memcpy(0x4317d94, 0x4317d94, 2)
==11033==    at 0x402879F: memcpy (in
/usr/lib/valgrind/vgpreload_memcheck-x86-linux.so)
==11033==    by 0x804979A: cstr_cat (tccpp.c:322)
==11033==    by 0x804B0A8: macro_is_equal (tccpp.c:1294)
==11033==    by 0x804BAA2: define_push (tccpp.c:1387)
==11033==    by 0x804BA5C: parse_define (tccpp.c:1604)
==11033==    by 0x804C28E: preprocess (tccpp.c:1792)
==11033==    by 0x804DD33: next_nomacro1 (tccpp.c:2649)
==11033==    by 0x804DAD4: next_nomacro_spc (tccpp.c:2986)
==11033==    by 0x804E90B: next_nomacro (tccpp.c:2993)
==11033==    by 0x804F874: next (tccpp.c:3525)
==11033==    by 0x805B832: decl0 (tccgen.c:6283)
==11033==    by 0x805BEC9: decl (tccgen.c:6497)
==11033==.