Bug 27838

Summary: Wrong code generated for for()-loop with enumerated type as index
Product: gcc Reporter: Oleh Derevenko <oder>
Component: c++Assignee: Not yet assigned to anyone <unassigned>
Status: RESOLVED DUPLICATE    
Severity: normal CC: adruab, Andreas.Glowatz, andreg, bobm75, boris, c.pop, cdfrey, charles, christophe.guillon, cxl, czang, d.bonekaemper, danfuzz, davids, davmac, devin, djk, dmeggy, duraid, evgeny, fabdouze, fm, gcc-bugs, gcc, ghouston, gino, gopalv82, grigory_zagorodnev, hayim, horst.lehser, hurbain, ja2morri, jason.elbaum, jfran, jochang, kalas, larschri, linuxadmin, lucho, l_heldt, mike, mor_gopal, noaml, oder, orion, pgonzalez, pierre.chatelier, pjh, puvar, rarob, renzo, rick.ju, spelis, steffen.zimmermann, steger, strasbur, sumii, thomas.anders, ulrich.lauther, zengpan, zybi
Priority: P3    
Version: 3.3.5   
Target Milestone: ---   
Host: x86 Target: x86
Build: x86 Known to work:
Known to fail: Last reconfirmed:

Description Oleh Derevenko 2006-05-31 12:51:39 UTC
I have enumerated type defined as follows

enum EQUEUEDWRITERFILEERROR
{
	QWFE__MIN,

	QWFE_REOPEN = QWFE__MIN,
	QWFE_WRITE,
	QWFE_OVERFLOW,

	QWFE__MAX,
};

For it I have prefix increment operator defined with macro
#define DEFINE_ENUM_INC_DEC(EnumType) \
	static inline EnumType &operator ++(EnumType &Value) { return (EnumType &)(++((int &)Value)); }

For following for-loop operator
	for (EQUEUEDWRITERFILEERROR qwfeError = QWFE__MIN; qwfeError != QWFE__MAX; ++qwfeError)

compiler generated wrong code for condition check
0x8090981 mov    0xffffffd7(%ebp),%edx
0x8090984 mov    0xffffffd7(%ebp),%al
0x8090987 inc    %edx
0x8090988 cmp    $0x3,%al
0x809098a mov    %edx,0xffffffd7(%ebp)
0x809098d jne    0x8090940 

That is, variable is incremented in parallel with loop condition check and result of increment has effect to loop condition only on the next pass. This causes the loop to execute one time more than required.

Compilation options used
-malign-double -fshort-enums -freg-struct-return -fno-exceptions -g -O3 -march=pentium -fno-rtti -fconserve-space
Comment 1 Andreas Schwab 2006-05-31 13:04:37 UTC
You are violating the aliasing rules.

*** This bug has been marked as a duplicate of 21920 ***